<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Galih Laras Prakoso on Medium]]></title>
        <description><![CDATA[Stories by Galih Laras Prakoso on Medium]]></description>
        <link>https://medium.com/@galihlprakoso?source=rss-98d80191145d------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*quFgcx8nR276qHjN5aj87Q.png</url>
            <title>Stories by Galih Laras Prakoso on Medium</title>
            <link>https://medium.com/@galihlprakoso?source=rss-98d80191145d------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 05:59:41 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@galihlprakoso/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Product Development: Premature Solution.]]></title>
            <link>https://medium.com/@galihlprakoso/product-development-premature-solution-4dcd0b36512d?source=rss-98d80191145d------2</link>
            <guid isPermaLink="false">https://medium.com/p/4dcd0b36512d</guid>
            <category><![CDATA[product]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[product-development]]></category>
            <dc:creator><![CDATA[Galih Laras Prakoso]]></dc:creator>
            <pubDate>Sat, 16 Nov 2024 02:51:21 GMT</pubDate>
            <atom:updated>2024-11-16T06:12:29.920Z</atom:updated>
            <content:encoded><![CDATA[<h4>Welcome to the Feature Factory hell, Enjoy!</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*uzRSsw4MSJz2Ektq.png" /><figcaption><a href="https://productcoalition.com/my-favorite-product-management-memes-and-the-kernel-of-truth-behind-them-2be1e9bdab01">https://productcoalition.com/my-favorite-product-management-memes-and-the-kernel-of-truth-behind-them-2be1e9bdab01</a></figcaption></figure><p>It has been five years since my first encounter with Product Development (PD). During this time, I’ve worked with various PD teams, each with its own character and unique approaches to product development. I’ve come to realize that some of my experiences in PD involved what I call <strong><em>premature solutions</em></strong>.</p><h3>What is a Premature Solution?</h3><p>A premature solution is one of the biggest challenges in PD, often leading to significant negative impacts on the team and, in severe cases, the entire company. For some unfortunate companies, delivering premature solutions becomes a cultural norm, dragging their PD teams into the abyss of a “<strong><em>Feature Factory</em></strong>.”</p><h3>What is Premature?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zVrMyBPVqjgVrYDXI1OKaQ.png" /></figure><p>According to Oxford Languages, “premature” means occurring or done before the usual or proper time — too early. The word is often associated with premature babies. For example, when a baby is born three weeks early, we say, “the baby was three weeks premature.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*mWM-R8nk3xDMMLpN" /><figcaption><a href="https://www.npr.org/2021/11/10/1054567559/boy-who-was-under-one-pound-at-birth-is-the-worlds-most-premature-infant-to-surv">https://www.npr.org/2021/11/10/1054567559/boy-who-was-under-one-pound-at-birth-is-the-worlds-most-premature-infant-to-surv</a> — Michelle Butler of Eutaw, Ala., and son Curtis Means in Birmingham, Ala. The child has been certified by Guinness World Records as the world’s most premature baby to survive.</figcaption></figure><p>Premature births can lead to serious health and developmental challenges. Babies born too early may miss critical stages of physical development, which can result in complications like neurodevelopmental impairments, vision and hearing issues, and even reduced survival rates.</p><p>This analogy highlights the importance of solution-design process in PD. Just as premature birth can hinder a baby’s growth, a premature solution in PD can cause serious problems for both the product and the company.</p><h3>What is Premature Solution in Product Development?</h3><p>Let’s explore this concept using the analogy of premature birth. A baby requires nine months of development to grow properly. In PD, while we’re not strictly talking about time, the solution development process follows a similar logic.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/626/0*lZ2IJ6Cc_TicTJW-.jpg" /><figcaption><a href="https://www.freepik.com/free-vector/hand-drawn-fetal-development-infographic_21077189.htm#fromView=keyword&amp;page=1&amp;position=2&amp;uuid=f8dfbded-c119-4e5d-924f-149779fdca4c">https://www.freepik.com/free-vector/hand-drawn-fetal-development-infographic_21077189.htm#fromView=keyword&amp;page=1&amp;position=2&amp;uuid=f8dfbded-c119-4e5d-924f-149779fdca4c</a></figcaption></figure><p>In this context (about Product Development), we’re not really talking about the time, because time might vary depends on a lot of factor, but I will talk more on the <strong><em>solution development process.</em></strong></p><h4>The Ideal Problem Solution Feedback Loop</h4><p>The ideal development process can be represented by this diagram:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GCQCrF77TYb641EqoLHGuQ.png" /><figcaption>Ideal Problem-Solution Feedback Loop Cycles</figcaption></figure><p>The problem flows from upstream to downstream teams, ensuring feedback is collected at every stage before a solution emerges. An ideal solution is born only when the full <em>Problem-Solution Feedback Loop</em> is completed.</p><p>Unfortunately, some teams operate differently. In these cases, the downstream teams, such as Engineering, often have no clue about the feature they’re building. They don’t understand its purpose, impact on customers, or how to measure its success. In the worst scenarios, even the Product team doesn’t know the “why” behind the features they’re creating.</p><h4>Premature Problem Solution Feedback Loop</h4><p>In a premature feedback loop, the problem doesn’t flow to the downstream teams. Instead, solutions are dictated by upstream, strategic, or business teams without sufficient context.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*cSBATJ33XphmG_gq6QLNgw.png" /><figcaption>Premature Problem-Solution Feedback Loop Cycles (Feature Factory Hell)</figcaption></figure><p>This process transforms PD teams into “<strong>Feature Factories</strong>,” where they develop features dictated by upstream demands without truly solving customer problems. They say “yes” to almost everything, leading to shallow solutions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*uzRSsw4MSJz2Ektq.png" /><figcaption><a href="https://productcoalition.com/my-favorite-product-management-memes-and-the-kernel-of-truth-behind-them-2be1e9bdab01">https://productcoalition.com/my-favorite-product-management-memes-and-the-kernel-of-truth-behind-them-2be1e9bdab01</a></figcaption></figure><p>There are exceptions to this rule. Some companies have upstream teams with strong product and technical expertise, akin to having a wise and intelligent dictator running a country. In such cases, even a “Feature Factory” can function well. But when this expertise is absent, the results can be disastrous.</p><h3>How to Avoid Becoming Feature Factory that Deliver Premature Solutions?</h3><p>First, assess the conditions in your company:</p><ul><li>Do upstream teams have strong product and technical knowledge?</li><li>Are you part of an exceptional team led by wise, intelligent leaders with strong product and technical knowledge?</li></ul><p>If yes, being a Feature Factory might not be a bad thing. However, for companies that lack these conditions (most companies), it’s crucial to ensure the <strong><em>Problem-Solution Feedback Loop</em></strong> is <strong>complete</strong>. Problems — and the “why” behind them — must flow to downstream teams. The Engineering team should fully understand the problem to propose solutions that fit customer needs while adhering to business constraints.</p><p>Let’s take a look at this illustration as an example!</p><h4>Premature Solution Example</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rJHZdjFifLqMwErpN7Jqew.png" /><figcaption>Generated by Microsoft Copilot</figcaption></figure><p><strong>Problem:</strong><br>Customers frequently complain that their orders on a food delivery app get delayed, but the app does not inform them about the delay in real time.</p><p><strong>Business Team’s Request:</strong><br>The business team requests the engineering team to add a feature that sends manual notifications to users about delays, allowing customer support representatives to send messages to affected users.</p><p>The engineering team, without fully understanding the problem, develops a manual notification system for customer support staff, which results in more work for support teams and slower responses for customers.</p><h4>Ideal Solution Example</h4><p><strong>Problem Analysis:</strong><br>The root problem is that delivery delays are caused by logistical bottlenecks such as restaurants taking longer to prepare food or drivers being unavailable. Customers want real-time updates and transparency, not just notifications about delays.</p><p><strong>Proposed Ideal Solution:</strong></p><ol><li><strong>Automated Delay Detection:</strong> Use machine learning or simple logic to detect delays by monitoring delivery milestones, such as order preparation time exceeding a threshold or driver assignment taking too long.</li><li><strong>Proactive Notifications:</strong> Automatically notify users about delays as soon as they are detected, explaining the cause (e.g., “The restaurant is preparing your order. We apologize for the delay.”).</li><li><strong>Driver Reassignment:</strong> If driver unavailability is the cause, introduce an automatic reassignment mechanism to ensure orders are delivered faster.</li><li><strong>Transparency Features:</strong> Add a live tracking feature with estimated preparation and delivery times so customers can independently monitor their orders.</li></ol><p>When problems are fully understood across all teams — from upstream to downstream — the resulting solution is more mature. It addresses the customer’s needs effectively without introducing significant product or technical debt.</p><h3>Conclusion</h3><p>Delivering premature solutions in Product Development can lead to a cascade of problems, turning teams into mere “Feature Factories” that miss the mark on solving real customer needs. To avoid this, it’s essential to complete the <strong>Problem-Solution Feedback Loop</strong>, ensuring every team — upstream and downstream — understands the “why” behind what they are building. Mature solutions emerge from collaboration, clear communication, and a deep understanding of the problem. When done right, this process not only delivers impactful products but also fosters a culture of innovation and purpose within the PD team.</p><p>If you found this article helpful, please don’t forget to give some claps (you can clap more than once!), leave your comments, and share this article with your friends! If you want to read more personal blog posts, please subscribe to my <a href="https://ghackdev.substack.com/">Substack</a> account! Thank you!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4dcd0b36512d" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Product Development: Build Your Product on a Solid Rock, Not Shifting Sands]]></title>
            <link>https://medium.com/@galihlprakoso/product-development-build-your-product-on-a-solid-rock-not-shifting-sands-f0c0b96fe5fb?source=rss-98d80191145d------2</link>
            <guid isPermaLink="false">https://medium.com/p/f0c0b96fe5fb</guid>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[product-manager]]></category>
            <category><![CDATA[product]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[product-development]]></category>
            <dc:creator><![CDATA[Galih Laras Prakoso]]></dc:creator>
            <pubDate>Wed, 06 Nov 2024 04:17:11 GMT</pubDate>
            <atom:updated>2024-11-06T04:22:10.613Z</atom:updated>
            <content:encoded><![CDATA[<h4>Rather than building your product on a fragile foundation, why not lay it on a solid one?</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1012/1*a3NtStPYL-fU2NwZDdh1jw.png" /></figure><p>Yesterday, I came across an interesting article from <a href="https://hbr.org/2024/11/how-to-avoid-the-agility-trap">HBR</a> that made me realize that all our efforts in building a great product could be completely wasted if we build it on the wrong foundation. By observing some product development teams in various companies struggling to deliver their products to customers, I noticed that many of these teams are trapped in the wrong practices of Agile. It’s as if they’re stuck in quicksand, sinking deeper and deeper. They’re building their products on “fragile” sands instead of solid rock.</p><h3>Strategic Constants vs Transient Factors</h3><p>Strategic Constants and Transient Factors. Strategic Constants are the core principles, values, and long-term objectives that form the foundation of a company’s product and vision. These constants are the unchanging aspects of a business strategy, like a clear focus on customer satisfaction, commitment to quality, or a unique mission that defines the brand. Companies that stay true to their strategic constants create a sense of stability and purpose, allowing them to build trust with customers and sustain growth over time.</p><p>On the other hand, Transient Factors are the temporary trends, market shifts, and short-term demands that can often be tempting distractions. These include rapid feature additions, integrations with the latest popular tools, or chasing fleeting customer preferences. While adapting to these factors can be beneficial, companies that focus too much on them risk losing sight of their core vision. They may end up investing in features or strategies that lack long-term value, ultimately leading to wasted effort and resources.</p><h3>Re-Thinking Agile</h3><p>Before we begin, let’s start by re-thinking the definition of Agility in our minds. We can do this by looking at these examples, grouped by their sectors. It may be a long list, but please skim through these examples quickly and observe the patterns that emerge (or you can skip the rest if you feel you’ve already grasped the pattern).</p><h4><strong>Entertainment and Media</strong></h4><ul><li><strong>Netflix</strong>: Netflix exemplifies strategic constancy by staying committed to delivering high-quality content. Even during its shift from DVD rentals to streaming, Netflix’s focus was on providing the best viewing experience, not on the delivery medium. This core strategy allowed Netflix to pivot effectively, continually adding original content and improving user experience as part of its long-term vision.</li><li><strong>Spotify</strong>: Spotify has focused on the constant demand for accessible and personalized music. By concentrating on algorithm-driven recommendations and an expansive music library, Spotify has consistently met listeners’ preferences, reinforcing its competitive advantage while adapting to technological advances and distribution channels.</li></ul><h4><strong>E-Commerce and Retail</strong></h4><ul><li><strong>Amazon</strong>: Amazon’s founder Jeff Bezos emphasized “strategic constants” like low prices, fast delivery, and vast selection. These constants guided its development, from creating fulfillment centers to offering Amazon Prime, ensuring Amazon’s growth regardless of market fluctuations.</li><li><strong>Costco</strong>: Costco is known for maintaining low prices through bulk purchasing and a limited SKU selection. This core strategy allows Costco to remain competitive even amid changing retail trends by offering high value and consistent customer satisfaction.</li></ul><h4><strong>Education and Learning</strong></h4><ul><li><strong>Duolingo</strong>: Duolingo has consistently focused on accessibility and gamification in language learning, which has remained relevant as it expanded. By keeping its offerings engaging and free, it appeals to a broad audience while adjusting features based on user feedback without compromising its core mission.</li><li><strong>Coursera</strong>: By focusing on accessible, high-quality learning experiences, Coursera has created a scalable model for online education. Its partnerships with universities and institutions highlight a commitment to affordable, flexible education, ensuring Coursera’s relevance amid changing learning demands.</li><li><strong>Khan Academy</strong>: Khan Academy’s dedication to providing free, high-quality education for everyone remains its constant. This unwavering mission has allowed it to grow with support from foundations and adapt as digital learning becomes mainstream.</li></ul><h4><strong>Financial Technology</strong></h4><ul><li><strong>PayPal</strong>: PayPal has anchored itself in the principle of secure and easy online transactions. Its adaptability in offering new services (like Buy Now, Pay Later) stems from its strategic constant of providing safe and convenient digital payments, which keeps it aligned with the evolving financial landscape.</li></ul><h4><strong>Consumer Technology</strong></h4><ul><li><strong>Apple</strong>: Apple’s focus on user-centric design and premium quality has been its constant. Rather than chasing trends, Apple has concentrated on building an ecosystem that enhances user experience and loyalty. This focus enables them to introduce products with enduring appeal, as seen with iPhones, iPads, and the Apple Watch.</li><li><strong>Samsung</strong>: Samsung’s consistent commitment to innovation and product quality — especially in screens and processors — has made it a leader in electronics. By investing heavily in R&amp;D and focusing on superior hardware, Samsung has maintained its position in a highly competitive market.</li></ul><h4><strong>Gaming</strong></h4><ul><li><strong>Nintendo</strong>: Nintendo’s focus on accessible and family-friendly gaming has allowed it to thrive even when competitors focus on graphics or online play. The constant appeal of franchises like Mario and Zelda and unique platforms like the Switch showcase Nintendo’s dedication to this long-term vision.</li><li><strong>Epic Games</strong>: With the Unreal Engine, Epic Games has focused on empowering game developers with versatile tools, providing value beyond just game titles. Fortnite’s success also stems from Epic’s strategic commitment to community-building, player engagement, and ongoing content updates.</li></ul><h4><strong>Health and Wellness</strong></h4><ul><li><strong>Johnson &amp; Johnson</strong>: In health and wellness, Johnson &amp; Johnson’s emphasis on reliability and trust has been a defining constant. Their broad portfolio in healthcare, coupled with a commitment to safety and innovation, has helped them maintain strong consumer and provider trust.</li><li><strong>Fitbit</strong>: Before being acquired by Google, Fitbit focused on wellness tracking and personal health data. This focus on making health tracking easy and accessible was a strategic constant that allowed Fitbit to gain a loyal user base, eventually leading to expansion into health analytics and wearable technology integration.</li></ul><h4><strong>Logistics and Supply Chain</strong></h4><ul><li><strong>DHL</strong>: DHL’s core focus on international logistics excellence and customer satisfaction has sustained its leadership. By optimizing supply chains and investing in technology to enhance tracking and delivery, DHL has adapted to global demands while keeping its strategic goals stable.</li><li><strong>FedEx</strong>: FedEx has continuously emphasized speed and reliability, making express shipping its strategic constant. This focus has guided investments in aircraft, logistics hubs, and automated technology to stay ahead in a competitive logistics industry.</li></ul><h4><strong>Automotive</strong></h4><ul><li><strong>Toyota</strong>: Toyota’s long-term commitment to quality, efficiency, and innovation (e.g., lean manufacturing) has made it resilient. Its approach to hybrid and electric vehicles reflects a strategic focus on reliability and sustainability, aligning with consumer expectations while exploring newer trends.</li><li><strong>Tesla</strong>: Tesla’s commitment to sustainable energy and innovation in electric vehicles has driven its success. Rather than focusing on short-term gains, Tesla anchors its strategy in advancing battery technology and sustainable energy infrastructure, allowing it to adapt within these constants as the EV market evolves.</li></ul><p>Let’s take a moment to reflect on how companies that focus on strategic constants, rather than transient factors, are able to create products that are loved by their customers. These companies can also maintain growth and sustainability.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/888/1*stO8Wr0WxhSwJbdVgl-EiA.jpeg" /><figcaption><a href="https://www.linkedin.com/pulse/get-comfy-pivots-emilee-cook-rxf2e/">https://www.linkedin.com/pulse/get-comfy-pivots-emilee-cook-rxf2e/</a></figcaption></figure><p>I’ve seen many product development teams struggle with the “endless” task of trying to adapt to market shifts. They continue to add more features to their products, integrate with more third parties, and focus solely on acquiring more users — while neglecting their strategic constants and getting distracted by the never-ending chase of transient factors.</p><p>I’m not trying to be religious (definitely not 🤣), but a teaching from Jesus, shared through a parable, is always on my mind when I talk about this. I believe it provides a simple and intuitive way to explain the concept of Strategic Constants and Transient Factors.</p><blockquote><strong>“Therefore everyone who hears these words of mine and puts them into practice is like a wise man who built his house on the rock. The rain came down, the streams rose, and the winds blew and beat against that house; yet it did not fall, because it had its foundation on the rock. But everyone who hears these words of mine and does not put them into practice is like a foolish man who built his house on sand. The rain came down, the streams rose, and the winds blew and beat against that house, and it fell with a great crash.”</strong></blockquote><blockquote>— Matthew 7:24–27 (NIV)</blockquote><p>This parable is having the same frequency about this trade-offs on the product development:</p><blockquote><strong>“everyone who focus more on building their foundation on the strategic constants is like a wise man who built his house on the rock, while everyone who focus more on building their foundation on the transient factors is like a foolish man who built his house on sand.”</strong></blockquote><h3><strong>Building Product on Top Of Sands</strong></h3><p>We can see how focusing on transient factors, rather than strategic constants, can lead to challenges. Startups like WeWork and another well-known case in the U.S. illustrate this issue. WeWork, for example, pursued rapid expansion with the ambition to redefine workspaces globally. However, its strategy was heavily reliant on constant growth and high valuations rather than a stable foundation in consistent demand or sustainable practices. When market conditions shifted, this approach left them vulnerable, demonstrating the risks of building on a foundation that isn’t rock-solid.</p><p>Similarly, another prominent startup in the U.S. faced difficulties by prioritizing rapid user acquisition and expanding services without establishing a sustainable business model. Like WeWork, this kind of company built its growth around changing market demands and investor expectations. When these external factors evolved, it struggled to adapt, lacking the stability that comes from having a more resilient core focus.</p><p>These examples highlight the importance of strategies grounded in long-term constants rather than quick wins. By building on strategic constants — such as a clear value proposition or reliable customer base — companies can maintain growth and resilience even when market conditions shift. This approach aligns with the idea of “building on top of a solid rock,” where a strong foundation supports not only survival but also sustainable success.</p><h3><strong>Building Product on Top Of a Solid Rock</strong></h3><p>Airbnb is a powerful example of building on “solid rock” through the use of “strategic constants” rather than “transient factors.” Just as in the parable where a wise man builds his house on rock to withstand storms, Airbnb anchored its business in a core principle: trust. From the start, Airbnb focused on building a reputation of safety and reliability by implementing reviews, safety measures, and customer support systems. Instead of only chasing rapid growth, Airbnb invested in this stable foundation of trust. When the pandemic challenged the travel industry, Airbnb was able to pivot to local travel and online experiences, leaning on its rock-solid commitment to quality and safety to stay relevant and resilient.</p><p>Slack also reflects the parable’s lesson by establishing a foundation on “strategic constants” rather than shifting with every trend. Rather than expanding in all directions or adding countless features, Slack focused on its main constant: improving team communication. This commitment to reliable, user-friendly communication tools allowed Slack to build a strong, stable user base that valued consistency and quality. Like the house built on rock, Slack’s product could stand firm even in a dynamic tech landscape, maintaining its relevance and value by sticking to its core purpose rather than chasing temporary trends.</p><p>Square, too, illustrates the wisdom of “building on rock” through its dedication to strategic constants. Instead of being swayed by trends, Square focused from the start on empowering small businesses — a purpose that has remained constant. Square started with a straightforward payment tool and gradually added products, but each addition was aligned with the core mission of supporting small businesses. This strong foundation has enabled Square to grow responsibly and weather market changes, demonstrating the strength and stability that come from building on rock, not sand.</p><h3><strong>The Wrong Definition of Agile</strong></h3><p>In the early stages of my journey in product development as a Software Engineer, I also carried the wrong definition of Agile. I thought Agile meant that we should adapt to ever-changing market conditions very quickly and be able to shift our direction accordingly. With this mindset, I found it normal when I saw many startups around me pivoting frequently, even if those pivots deviated from their initial vision. However, at the time, something about this didn’t sit right with me, and my intuition told me that it shouldn’t be this way.</p><p>After several years in product development, I realized that Agile doesn’t mean this at all. It aligns more with the concept I encountered while learning about Buddhism, which emphasizes the teaching of “The Middle Way.” Instead of being extreme in either direction, we should approach Agile with a balanced perspective. It’s essential to be adaptive to market conditions, but it’s equally important to establish our core foundation on solid rock, not shifting sands.</p><h3><strong>Conclusion</strong></h3><p>In conclusion, the key to building a successful and sustainable product lies not in constantly chasing transient trends or adapting to every shift in the market, but in establishing a solid foundation rooted in strategic constants. The stories of companies like Netflix, Airbnb, and Square illustrate the power of maintaining a clear, unwavering focus on core principles, even in the face of adversity or changing market conditions. These companies have succeeded not by being reactive to every fluctuation but by remaining true to their long-term vision and adapting in ways that reinforce their foundation.</p><p>On the other hand, companies that focus solely on transient factors — such as rapid growth, flashy features, or chasing the latest trends — risk building their success on unstable ground. The failures of companies like WeWork demonstrate the dangers of neglecting the importance of strategic constancy, leaving them vulnerable when market forces inevitably shift.</p><p>As product development teams, it’s essential to embrace agility in a balanced way. Agile methods should enable teams to adapt and innovate, but always with a strong commitment to the long-term vision and core values. Just as a house built on sand will collapse under pressure, a product built on fleeting trends will not withstand the test of time. Instead, we must build our products on rock-solid foundations — just like the wise man in the parable — ensuring that our product’s core purpose remains steadfast, no matter what external forces we encounter.</p><p>If you found this article helpful, please don’t forget to give some claps (you can clap more than once!), leave your comments, and share this article with your friends! If you want to read more personal blog posts, please subscribe to my <a href="https://ghackdev.substack.com/">Substack</a> account! Thank you!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f0c0b96fe5fb" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Product Development: Garbage In, Garbage Out]]></title>
            <link>https://medium.com/@galihlprakoso/product-development-garbage-in-garbage-out-ab3d8dd2b5c9?source=rss-98d80191145d------2</link>
            <guid isPermaLink="false">https://medium.com/p/ab3d8dd2b5c9</guid>
            <category><![CDATA[product]]></category>
            <category><![CDATA[product-design]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[tech]]></category>
            <category><![CDATA[product-management]]></category>
            <dc:creator><![CDATA[Galih Laras Prakoso]]></dc:creator>
            <pubDate>Sat, 19 Oct 2024 03:13:27 GMT</pubDate>
            <atom:updated>2024-10-19T11:21:47.281Z</atom:updated>
            <content:encoded><![CDATA[<h4>Garbage In, Garbage Out: Insanity is doing the same thing over and over again and expecting different results. — Albert Einstein.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-l2BX97gvzVHVEEe5SIAQw.png" /></figure><p>My journey in product development has taken me through various experiences, from working at multiple companies and part-time gigs with founders to launching my own products and companies. After all these “hell of a ride” experiences, coupled with insightful discussions with some amazing people, I’ve come to a clear realization: the essence of product development is people.</p><p>It’s not about the code, the interface, or even the strategy itself — it’s about the people behind every stage of the process. The fundamental truth is that the quality of the output — the product — relies entirely on the quality of the input: the people. This understanding narrows down the focus to what matters most in product development: cultivating a team of capable, committed, and growth-oriented individuals.</p><h3>The True Product is People</h3><p>When we talk about product development, it may seem like we’re discussing software, applications, or physical products. But at its core, product development is an ecosystem where <strong>people are the main drivers</strong>. Any product, whether a digital tool or a service, is ultimately shaped by the <strong>people</strong> who create, design, and refine it. It’s a cycle where problems or needs arise from people (customers), are addressed by people (the product development team), and result in solutions that go back to people (customers).</p><p>This can be visualized in an input-output flow:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6fcUFJQEvyKTpQI4WhTXSw.png" /><figcaption>Customer (Problems, Needs) → Product Development Team (Ideas → Design → Development) → Back to Customer.</figcaption></figure><p>In this flow, what happens in the middle — how the product development team handles the feedback, innovates solutions, and refines the product — ultimately determines the quality of the outcome. The quality of the “input” at each step directly impacts the “output” that reaches the customers. It’s essentially a cycle where everything starts and ends with <strong>people</strong>, just like a Yin and Yang relationship between <strong>customers</strong> and the <strong>product development team</strong>.</p><h3>The River Analogy: A Circular Product Development Process</h3><p>Imagine the product development process as a river. Picture a circular river that flows continuously, where each stage represents different contributors along the riverside. Customers, Boards and Executives, VPs, Managers, Designers, Developers, QA testers — they all live along the river, and their actions affect the water quality. If any of these people throws garbage into the river, the downstream stages will feel the consequences. If no one cleans it up, the garbage eventually reaches the customer, the very end of the flow, causing dissatisfaction or failure of the product.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*zvwRU5JrMhLptSwMCje9-A.png" /><figcaption>The River of Product Development</figcaption></figure><p>This analogy reflects a harsh reality I witnessed growing up by the riverside. Despite my efforts to educate people around me about not polluting the river, it stayed dirty because others upstream continued throwing waste. Similarly, in product development, no matter how diligent some teams are, if others upstream are careless or indifferent, the end result will suffer, In the context of a circular river, one person’s actions affect everyone downstream, and without a clear “upstream,” accountability becomes a <strong>shared responsibility </strong>for the <strong>entire team</strong>.</p><h3>Good Culture = Good Output</h3><p>Just as keeping a river clean requires the right behaviors and values, creating a great product requires fostering a healthy company culture. When a team operates in an environment where growth, responsibility, and open-mindedness are valued, they are more likely to produce high-quality results. It’s not enough to have smart or highly educated individuals on the team; what truly matters is their mindset, values, and willingness to commit to the vision.</p><p>If some members lack the right attitude, the entire flow suffers. This is why it’s essential to hire people who are not only skilled but also possess the right cultural fit and growth potential. As a famous quote goes, “<strong>Culture eats strategy for breakfast.</strong>” Culture is the unseen force that shapes behaviors and attitudes, ultimately determining the quality of the products produced.</p><h3>The Role of Leadership: A Servant Leader</h3><p>Cleaning up the metaphorical river requires strong leadership. Effective leaders are <strong>not</strong> “<strong>bosses</strong>” who issue commands from the comfort of their offices; they are “<strong>servant leaders</strong>” who roll up their sleeves and work alongside their teams to clean up the mess. They guide their teams by <strong>example</strong>, demonstrating the <strong>commitment</strong> needed to achieve a <strong>shared goal</strong>.</p><p>True leadership is about leveling up the people around you — helping them grow from a “5” to a “7” or from a “6” to a “9” on a scale of ten. It’s about fostering a sense of ownership and purpose among team members, aligning their individual goals with the company’s vision. A servant leader builds a <strong>culture</strong> where <strong>rewards</strong> are earned, and the consequences of <strong>poor performance</strong> are <strong>clearly understood</strong>. They set the tone for maintaining <strong>high standards</strong> and ensuring everyone on the team is <strong>aligned</strong>.</p><h3>When Business Can’t Wait: The Harsh Reality</h3><p>While it’s <strong>crucial</strong> to invest in <strong>people’s growth</strong>, business <strong>realities</strong> sometimes demand <strong>swift action</strong>. There will be moments when certain individuals are <strong>too slow</strong> to adapt or are <strong>unable to change.</strong> Just as a <strong>single person</strong> who continues to <strong>pollute</strong> the river can affect the <strong>entire ecosystem</strong>, an underperforming or <strong>uncooperative</strong> team member can <strong>drag down</strong> the team’s output.</p><p>The <strong>difficult truth</strong> is that not everyone can grow fast enough to keep up with the demands of a business. Leaders must <strong>recognize</strong> when to <strong>let go</strong> of individuals who are <strong>hindering progress</strong>. Even in nature, harsh decisions are sometimes made for the sake of evolution — such as when a bird lets a weak chick fall, ensuring that only the strongest survive. It may seem ruthless, but it’s necessary for the health of the team and the <strong>quality of the product</strong>.</p><h3>Small Teams: Less Garbage, Easier Cleanup</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YFdOL4iTFmrNGywmycLbEQ.png" /></figure><p>This is why companies often prefer <strong>smaller</strong>, <strong>highly skilled</strong> with <strong>growth-mindset</strong> kind of teams. A small group of dedicated, talented people is <strong>easier to manage</strong> and can maintain <strong>higher standards</strong>. For instance, companies like <strong>Valve</strong> and <strong>WhatsApp</strong> achieved remarkable results with <strong>small teams</strong> because they could focus on quality and foster a <strong>strong</strong>, <strong>shared culture</strong>. When there are fewer people involved, it’s easier to ensure everyone is aligned and <strong>committed</strong> to the <strong>same goals</strong>.</p><h3>Ensuring a Clean River: Practical Steps</h3><p>To enhance the product development process, companies should focus on:</p><ol><li><strong>Hiring for Cultural Fit:</strong> Skills can be taught, but attitudes are harder to change. Look for people who are aligned with the company’s values and growth mindset.</li><li><strong>Establishing Clear Rewards and Consequences:</strong> Just as there are laws against polluting rivers, set clear guidelines for performance, with appropriate rewards for achievement and consequences for poor performance.</li><li><strong>Investing in Leadership Development:</strong> Train leaders to be servant-leaders who guide by example and empower their teams to take ownership of the product.</li><li><strong>Listening to the Right Customers:</strong> Use feedback selectively. As noted in some famous books: “the loudest customers sometimes represent a small amount of real customers.” Identify customers who best represent your target audience to guide product development effectively.</li><li><strong>Maintaining a Small, Focused Team:</strong> Smaller teams often produce higher-quality work due to easier communication and stronger cultural alignment.</li></ol><h3>Conclusion</h3><p>In product development, the principle of “<strong>Garbage In, Garbage Out</strong>” holds true. The quality of the people — whether they are <strong>customers</strong> providing feedback or <strong>team members</strong> executing the work — determines the <strong>quality of the product</strong>. It’s about ensuring that every part of the process is <strong>nurtured</strong> and <strong>cared</strong> for, from customer input to leadership guidance, with a constant focus on <strong>keeping the “river” clean</strong>. Only then can we deliver products that truly meet and exceed the needs of the customers.</p><p>If you found this article helpful, please don’t forget to give some claps (you can clap more than once!), leave your comments, and share this article with your friends! If you want to read more personal blog posts, please subscribe to my <a href="https://ghackdev.substack.com/">Substack</a> account! Thank you!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ab3d8dd2b5c9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Rapid Development, Reliable Consistency: The Power of GraphQL Federation in Fintech]]></title>
            <link>https://medium.com/@galihlprakoso/rapid-development-reliable-consistency-the-power-of-graphql-federation-in-fintech-b6c40e199c97?source=rss-98d80191145d------2</link>
            <guid isPermaLink="false">https://medium.com/p/b6c40e199c97</guid>
            <category><![CDATA[fintech]]></category>
            <category><![CDATA[spring-boot]]></category>
            <category><![CDATA[graphql]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[Galih Laras Prakoso]]></dc:creator>
            <pubDate>Sun, 29 Sep 2024 07:11:27 GMT</pubDate>
            <atom:updated>2024-09-29T07:45:27.226Z</atom:updated>
            <content:encoded><![CDATA[<h4>Hello Hackers! In this blog post, we’ll explore how to leverage GraphQL Federation to design a robust system architecture that maximizes consistency and minimizes communication overhead, ultimately accelerating your development speed.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*n0SdluA2-sNQta7y4NEzdg.png" /><figcaption>Generated by DALL·E</figcaption></figure><p>In the competitive world of fintech, companies must operate with extreme efficiency, speed, and consistency. Microservices architectures are the backbone of modern fintech systems, but managing multiple services can introduce significant challenges related to communication, schema consistency, and integration. This is where GraphQL federation comes into play, offering a powerful solution to orchestrate data across distributed services while maintaining agility and accuracy.</p><p>In this article, we’ll explore why GraphQL federation is ideal for fintech companies and how it can revolutionize system design by addressing key challenges. Let’s dive into the primary reasons fintech organizations should adopt GraphQL federation for their system design.</p><h4>1. Schema Registry: Ensuring Consistency Across Microservices</h4><p>One of the fundamental challenges in any microservices architecture is ensuring consistency across the services. In fintech, where accuracy and compliance are paramount, having an inconsistent schema can lead to bugs, communication issues, and potential financial loss.</p><p>GraphQL federation uses a <strong>schema registry</strong>, ensuring that all teams working on different microservices operate with a consistent schema. This avoids confusion between services when schemas change, and the schema registry automatically communicates these changes to relevant services. For example, when a field is deprecated, all services that depend on it are notified through the schema registry. This eliminates the need for manual communication, thus reducing the likelihood of errors due to miscommunication or schema mismatches.</p><p>Additionally, maintaining consistency in the schema enables <strong>predictable behavior</strong> when scaling your microservices architecture, allowing individual teams to independently work on their services while ensuring integration is smooth and consistent.</p><h4>2. Boosting Development Speed in a Red Ocean Market</h4><p>Fintech companies often operate in a <strong>red ocean</strong> market, where the competition is fierce, and development speed is critical to stay ahead. By using GraphQL federation, teams can rapidly develop and deploy new features without worrying about schema inconsistencies or communication overhead.</p><p>A consistent schema maintained by the registry means developers don’t need to repeatedly discuss how new fields or entities should be represented. With the schema handled by GraphQL federation, teams can focus on developing their components without constantly synchronizing their models with other teams. This dramatically improves development speed while preserving quality and reducing technical debt.</p><p>Moreover, the <strong>agility</strong> provided by GraphQL allows developers to update and release features faster without causing disruption to other services. The whole system becomes more modular and streamlined, making it easier to pivot or iterate when necessary.</p><h4>3. Auto-Documentation and Simplified Client Queries</h4><p>Without a schema registry, documenting APIs and keeping that documentation up to date is a labor-intensive task, especially when dealing with numerous microservices. In fintech, where compliance, transparency, and clear communication with external stakeholders are essential, manual documentation can be prone to errors.</p><p>GraphQL federation simplifies this through <strong>auto-documentation</strong>. The schema registry acts as a source of truth, and clients can query the available schema to understand what data they can access. Clients no longer need to rely on individual microservices to implement custom endpoints or documentation tailored to their needs. Instead, the federation structure allows clients to request only the data they require, reducing communication costs and ensuring scalability.</p><p>This also helps in <strong>client-agnostic development</strong>: microservices teams don’t have to build specific endpoints for each type of client (whether internal or third-party), reducing overhead while offering more flexibility to clients.</p><h4>4. Handling Third-Party Integrations with Ease</h4><p>Fintech companies frequently integrate with third-party services such as banks, payment gateways, or regulatory bodies. Each integration introduces a risk of schema mismatch or inconsistency. Ensuring that your internal data models align with third-party schemas can be time-consuming and error-prone.</p><p>By leveraging GraphQL federation’s <strong>schema registry</strong>, the process of integrating third parties becomes more efficient. Any new integration can be mapped directly to your consistent internal schema, and updates to third-party schemas can be synchronized seamlessly. This ensures that internal teams and external partners are always aligned on how data should be represented and exchanged.</p><p>Furthermore, the centralized schema ensures that <strong>all microservices follow the same rules</strong> when interacting with external partners, reducing the chances of discrepancies across services.</p><h4>5. Orchestrating Efficient Data Delivery</h4><p>One of the critical pain points in traditional microservices architecture is orchestrating data retrieval efficiently. Regular RESTful microservices often result in over-fetching or under-fetching data, which can be costly in terms of performance, especially when dealing with financial transactions or large datasets.</p><p>GraphQL federation addresses this by allowing <strong>clients to request exactly the data they need</strong>, and not a byte more. By utilizing GraphQL resolvers, the system automatically orchestrates data retrieval across multiple microservices, ensuring efficient delivery without additional network requests or excessive processing. This results in <strong>better performance</strong> and more efficient use of system resources, which is critical for a fintech company dealing with high-traffic, real-time applications.</p><h4>6. Event-Driven Architecture in GraphQL Federation</h4><p>In a fintech environment, many processes are <strong>event-driven</strong>. For example, a new transaction might trigger several downstream processes, from updating account balances to sending notifications to different services. Handling these events in a traditional microservices architecture can lead to complex, tightly coupled systems.</p><p>GraphQL federation can make <strong>event-driven architectures</strong> easier by enabling microservices to respond to events in a loosely coupled manner. Services can publish and subscribe to events while using GraphQL as a unified API layer. The federation architecture allows events to propagate through different services without manual orchestration, using GraphQL subscriptions or other mechanisms to ensure that only the relevant data is requested and processed.</p><p>For instance, if a transaction triggers an event, GraphQL can ensure that each service involved only processes the specific data it needs to handle that event. This creates a more <strong>scalable</strong> and maintainable system, reducing the complexity of managing event-driven workflows across multiple services.</p><h4>7. Security and Compliance</h4><p>Security is a top concern for any fintech organization, especially when dealing with sensitive financial data. GraphQL federation helps enforce <strong>consistent security policies</strong> across the entire system, as access control can be managed centrally at the schema level. You can define authorization rules within your schema, ensuring that only authorized clients or services can access specific fields or operations.</p><p>This consistency also aids in <strong>regulatory compliance</strong>, as fintech companies need to adhere to strict data protection regulations such as GDPR, PCI-DSS, or SOC 2. By maintaining a single source of truth for data access and ensuring that all services follow the same rules, GraphQL federation simplifies audits and ensures compliance across your architecture.</p><h3>System Design</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bAy_xlYHlPg3F5P9NF6I_g.png" /><figcaption>System Design that utilize the concept of GraphQL Federation for Fintech Company</figcaption></figure><h4>Key Aspects of the Architecture:</h4><p><strong>1. GraphQL Gateway &amp; Schema Registry</strong>:</p><ul><li>The GraphQL API Gateway is responsible for handling client requests. The <strong>Schema Registry</strong> ensures consistency across the different Domain Graph Services (DGS), centralizing schema management and validation for microservices.</li></ul><p><strong>2. Domain Graph Services (DGS)</strong>:</p><ul><li><a href="https://netflix.github.io/dgs/">DGS</a> is the core of Netflix’s architecture, where each microservice (Accounts, Transactions, etc.) owns its respective domain and schema. This distributed ownership enables scaling, development autonomy, and schema validation within each DGS.</li></ul><p><strong>3. Event-Driven Architecture</strong>:</p><ul><li>The Event Bus handles event-driven services like balance updates and fraud detection. DGS microservices (such as <strong>Transactions</strong> and <strong>Payments</strong>) publish events to this bus.</li></ul><p><strong>4. Third-Party Integrations</strong>:</p><ul><li>Each integration (like <strong>Bank APIs</strong> or <strong>Payment Gateways</strong>) is managed via its own DGS to ensure schema consistency across internal and external systems.</li></ul><p><strong>5. Security and Compliance</strong>:</p><ul><li><strong>Authentication</strong> and <strong>Authorization</strong> are enforced centrally to ensure that sensitive fintech data is managed securely. This layer also handles <strong>Compliance Audits</strong> to meet regulatory requirements.</li></ul><p><strong>6. Client Flexibility</strong>:</p><ul><li>Various clients (Mobile, Web, Admin) can request exactly what they need via GraphQL, reducing data over-fetching and boosting performance.</li></ul><h3>Let’s Practice!</h3><p>Here’s a practical guide (simplified) to setting up GraphQL Federation for a fintech company using Java Spring Boot with Netflix’s DGS Framework. We’ll create:</p><p><strong>1. Domain Graph Services (DGS)</strong>:</p><ul><li><strong>Account Service</strong>: Manages customer accounts.</li><li><strong>Transaction Service</strong>: Handles transaction data.</li></ul><p><strong>2. GraphQL Gateway</strong>: Federates the schemas and handles client requests.</p><h4>Account Service</h4><p>You can initialize a Spring Boot application for account service by visiting this <a href="https://start.spring.io/">link</a>. And please follow this following configuration:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sAxqEXaM4Tu8BFLpJ_KnRA.png" /><figcaption>Spring Initializr — Account Service</figcaption></figure><p>Open the project using your favourite IDE, I recommend using <a href="https://www.jetbrains.com/idea/">IntellijIDEA</a> .</p><p>Create a schema file inside the resources/schema folder with name schema.graphqls .</p><pre>type Account @key(fields: &quot;id&quot;) {<br>    id: ID!<br>    name: String<br>    balance: Float<br>}<br><br>type Query {<br>    accounts: [Account]<br>}</pre><p>Edit the schema paths in the build.gradle file, point it to the schema folder where we place our GraphQL schema files:</p><pre>generateJava {<br> // point it to schema folder<br> schemaPaths = [&quot;${projectDir}/src/main/resources/schema&quot;]<br> packageName = &#39;com.ghackdev.accountservice.codegen&#39;<br> generateClient = true<br>}</pre><p>Run the generateJava gradle task:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aorOAcrsoms7KOzxGNoLRw.png" /></figure><p>Create new package with name datafetchers inside your root application package directory and create new Kotlin class with name AccountsDataFetcher inside that package:</p><pre>package com.ghackdev.accountservice.datafetchers<br><br>import com.ghackdev.accountservice.codegen.DgsConstants<br>import com.ghackdev.accountservice.codegen.types.Account<br>import com.netflix.graphql.dgs.DgsComponent<br>import com.netflix.graphql.dgs.DgsEntityFetcher<br>import com.netflix.graphql.dgs.DgsQuery<br><br>@DgsComponent<br>class AccountsDataFetcher {<br>    private val accounts: List&lt;Account&gt; = java.util.List.of(<br>        Account(&quot;1&quot;, &quot;John Doe&quot;, 1000.0),<br>        Account(&quot;2&quot;, &quot;Jane Smith&quot;, 2500.5)<br>    )<br><br>    @DgsQuery<br>    fun accounts(): List&lt;Account&gt; {<br>        return accounts<br>    }<br><br>    @DgsEntityFetcher(name = DgsConstants.ACCOUNT.TYPE_NAME)<br>    fun account(values: Map&lt;String?, Any?&gt;): Account {<br>        return accounts.stream()<br>            .filter { account: Account -&gt; account.id.equals(values[&quot;id&quot;]) }<br>            .findFirst()<br>            .orElse(null)<br>    }<br>}</pre><p>Use the generated DgsConstants and Account type files.</p><p>Replace the application.properties file inside the resources folder with this application.yml file:</p><pre>server:<br>  port: 8080<br><br>spring:<br>  application:<br>    name: account-service</pre><h4>Transaction Service</h4><p>You can initiate the Transaction Service similar to the way we generate the Account Service in the preceding section.</p><p>Create the schema for Transaction Service as follows:</p><pre>type Transaction {<br>    id: ID!<br>    accountId: ID!<br>    amount: Float<br>}<br><br>type Query {<br>    transactions: [Transaction]<br>}<br><br>type Account @key(fields: &quot;id&quot;) @extends {<br>    id: ID! @external<br>    transactions: [Transaction]<br>}</pre><p>And don’t forget to generate the types by using the generateJava Gradle task, and reate the TransactionDataFetcher class:</p><pre>import com.ghackdev.transactionservice.codegen.DgsConstants<br>import com.ghackdev.transactionservice.codegen.types.Account<br>import com.ghackdev.transactionservice.codegen.types.Transaction<br>import com.netflix.graphql.dgs.*<br>import java.util.stream.Collectors<br><br><br>@DgsComponent<br>class TransactionsDataFetcher {<br>    private val transactions: List&lt;Transaction&gt; = listOf(<br>        Transaction(&quot;1&quot;, &quot;1&quot;, 150.0),<br>        Transaction(&quot;2&quot;, &quot;1&quot;, 200.0),<br>        Transaction(&quot;3&quot;, &quot;2&quot;, 500.5)<br>    )<br><br>    @DgsEntityFetcher(name = DgsConstants.ACCOUNT.TYPE_NAME)<br>    fun account(values: Map&lt;String?, Any?&gt;): Account {<br>        return Account(values[&quot;id&quot;] as String, null)<br>    }<br><br>    @DgsQuery<br>    fun transactions(): List&lt;Transaction&gt; {<br>        return transactions<br>    }<br><br>    @DgsData(parentType = DgsConstants.ACCOUNT.TYPE_NAME, field = DgsConstants.ACCOUNT.Transactions)<br>    fun transactionsForAccount(dfe: DgsDataFetchingEnvironment): List&lt;Transaction&gt; {<br>        val account = dfe.getSource&lt;Account&gt;()<br>        if (account != null) {<br>            return transactions.stream()<br>                .filter { transaction: Transaction -&gt; transaction.accountId.equals(account.id) }<br>                .collect(Collectors.toList())<br>        }<br><br>        return emptyList()<br>    }<br>}</pre><p>Replace the application.properties file inside the resources folder with this application.yml file:</p><pre>server:<br>  port: 8081<br><br>spring:<br>  application:<br>    name: transaction-service</pre><h4>GraphQL Gateway</h4><p>For the GraphQL gateway, we’re going to use simple NodeJS server that uses some Apollo libraries:</p><ul><li><a href="https://github.com/apollographql/apollo-server">@apollo/server</a></li><li><a href="https://www.npmjs.com/package/@apollo/gateway">@apollo/gateway</a></li></ul><p>Initate new NodeJS project by using this package.json file:</p><pre>{<br>  &quot;name&quot;: &quot;apollo-gateway&quot;,<br>  &quot;version&quot;: &quot;1.0.0&quot;,<br>  &quot;description&quot;: &quot;&quot;,<br>  &quot;main&quot;: &quot;index.js&quot;,<br>  &quot;scripts&quot;: {<br>    &quot;test&quot;: &quot;echo \&quot;Error: no test specified\&quot; &amp;&amp; exit 1&quot;,<br>    &quot;start&quot;: &quot; node index.js&quot;<br>  },<br>  &quot;keywords&quot;: [],<br>  &quot;author&quot;: &quot;&quot;,<br>  &quot;license&quot;: &quot;ISC&quot;,<br>  &quot;dependencies&quot;: {<br>    &quot;@apollo/gateway&quot;: &quot;^2.0.3&quot;,<br>    &quot;@apollo/server&quot;: &quot;^4.11.0&quot;,<br>    &quot;graphql&quot;: &quot;^16.5.0&quot;<br>  }<br>}</pre><p>Create new file with name index.js in the root directory:</p><pre>const { ApolloServer } = require(&#39;@apollo/server&#39;);<br>const { startStandaloneServer } = require(&#39;@apollo/server/standalone&#39;);<br>const {ApolloGateway, IntrospectAndCompose} = require(&#39;@apollo/gateway&#39;)<br><br>const gateway = new ApolloGateway({<br>    supergraphSdl: new IntrospectAndCompose({<br>        subgraphs: [<br>            { name: &#39;accounts&#39;, url: &#39;http://localhost:8080/graphql&#39; },<br>            { name: &#39;transactions&#39;, url: &#39;http://localhost:8081/graphql&#39; },<br>        ]<br>    })<br>});<br><br>const server = new ApolloServer({ gateway, subscriptions:false, tracing:true });<br><br>startStandaloneServer(server).then(({url}) =&gt; {<br>  console.log(`🚀  Server ready at ${url}`);<br>});</pre><p>You will need to define the subgraphs URLs that points to our DGS services.</p><h4>Testing!</h4><p>Now you can test it sending request to the Gateway Service:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0w1jYq1OfQ32NOAjDJpPQA.png" /><figcaption>Insomnia — GraphQL query to Gateway Service</figcaption></figure><p>You can see the full version of the code <a href="https://github.com/galihlprakoso/graphql-federation-for-fintech-example">here</a>.</p><h3>Conclusion</h3><p>Implementing GraphQL Federation with the Netflix DGS framework offers significant advantages for fintech companies navigating a complex microservices landscape.</p><ol><li><strong>Consistency</strong>: Schema registries ensure all services adhere to a unified schema, reducing errors and miscommunication.</li><li><strong>Development Speed</strong>: Teams can work independently within their domains, accelerating feature delivery and fostering innovation.</li><li><strong>Client Simplicity</strong>: A single GraphQL Gateway simplifies client interactions, enabling seamless querying of multiple services while reducing backend complexity.</li><li><strong>Scalability</strong>: The federated architecture allows for easy scaling of individual services, adapting to growing demands without compromising performance.</li><li><strong>Efficient Integration</strong>: A consistent schema facilitates straightforward integration with third-party services, essential for delivering comprehensive fintech solutions.</li></ol><p>In summary, GraphQL Federation enhances collaboration and operational efficiency, empowering fintech companies to thrive in a competitive environment while setting the stage for future growth.</p><p>If you found this article helpful, please don’t forget to give some claps (you can clap more than once!), leave your comments, and share this article with your friends! If you want to read more personal blog posts, please subscribe to my <a href="https://ghackdev.substack.com/">Substack</a> account! Thank you!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=b6c40e199c97" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Rapidly Build a Powerful RAG Application with Django, LangChain, React, and ElasticSearch]]></title>
            <link>https://medium.com/@galihlprakoso/rapidly-build-a-powerful-rag-application-with-django-langchain-react-and-elasticsearch-efcae57755de?source=rss-98d80191145d------2</link>
            <guid isPermaLink="false">https://medium.com/p/efcae57755de</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[llm]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[Galih Laras Prakoso]]></dc:creator>
            <pubDate>Sat, 21 Sep 2024 04:30:14 GMT</pubDate>
            <atom:updated>2024-09-21T04:30:14.484Z</atom:updated>
            <content:encoded><![CDATA[<h4>Hello, Hackers! In this article, I will share my approach to quickly building a simple yet powerful RAG application.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*M9YfKskpUFquKTrIqR215Q.jpeg" /><figcaption>Photo by <a href="https://unsplash.com/@wocintechchat?utm_content=creditCopyText&amp;utm_medium=referral&amp;utm_source=unsplash">Christina @ wocintechchat.com</a> on <a href="https://unsplash.com/photos/person-typing-on-laptop-UIqpEPaiHQM?utm_content=creditCopyText&amp;utm_medium=referral&amp;utm_source=unsplash">Unsplash</a></figcaption></figure><p>RAG (Retrieval Augmented Generation) is an architectural approach designed to enhance the capability of LLMs by enabling them to answer queries based on provided documents. In simple terms, it focuses on retrieving relevant documents for a query and providing them as context to the LLM.</p><p>The application of RAG can help solve many real-world problems. For instance, I’ve used RAG to assist a friend with an organizational issue, specifically utilizing it to search for specific rules or policies within a collection of PDFs, each containing hundreds of pages that could take hours to look for specific rule or policy inside the documents.</p><h3>High Overview</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pW8zirw0dBLjoeGsl10VZw.png" /><figcaption>RAG Application Architecture</figcaption></figure><p>This is a simple RAG application architecture. Let’s break the architecture down into three main processes:</p><ul><li><strong>Document Embedding and Indexing in a Vector Database</strong>: In this step, we simply convert the content of the documents into embeddings using an Embedding Model and store the document embeddings in a Vector Database.</li><li><strong>Document Retrieval from the Vector Database</strong>: We convert the user’s query into an embedding and use it to retrieve relevant documents from the Vector Database.</li><li><strong>Querying the LLM with Retrieved Documents as Context</strong>: The query is sent to the LLM with the retrieved relevant documents as context, enabling the LLM to generate an accurate response.</li></ul><h3>Let’s Build!</h3><p>In this article, I want to show you how to quickly build a simple RAG application in practice! I will provide a step-by-step guide and outline the tech stack used to build each component of your RAG application.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CUCe_cz6Cu6Fkq2ulp7wpA.png" /><figcaption>System Design of RAG Application</figcaption></figure><p>We will go over each component in the system design one by one, and I will share the implementation details of each component.</p><h3>ERD</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GMarNfYftKTyn88-iPx2VQ.png" /><figcaption>RAG Application ERD</figcaption></figure><p>This simple ERD will be sufficient to build our basic RAG application:</p><ul><li><strong>Session</strong>: Manages chat sessions, with a ‘title’ field that contains the LLM-generated title based on the user’s first message.</li><li><strong>Chat</strong>: Manages chats within each session. This model includes two important fields: ‘role’ (to indicate whether the message is from the user or the AI) and ‘content’ (which contains the chat content in markdown format).</li><li><strong>Document</strong>: Manages all documents that will be retrieved and used as context for the LLM model.</li><li><strong>Page</strong>: Documents will be split into pages. Instead of embedding the entire document — which isn’t feasible for large documents — we chunk the document into pages. This way, the AI refers to pages from multiple documents as context, rather than entire documents.</li></ul><h3>React</h3><p>On the frontend side, I chose to use <a href="https://react.dev/">React</a> for its simplicity and efficiency. The main reason is that I found a very useful library by Ant Design, which can be quickly utilized with minimal setup, configuration, and customization.</p><h4>@ant-design/pro-chat</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7K0ctQD6eOqjifmKau0jYg.png" /><figcaption>@ant-design/pro-chat</figcaption></figure><p>This library provides nearly all the features needed to cover the functionalities required for an LLM chat interface. There’s no need to build everything from scratch — check it out <a href="https://github.com/ant-design/pro-chat">here</a>. Additionally, you can use the <a href="https://ant.design/docs/react/introduce">Ant Design UI library</a> to swiftly build extra UI components for your RAG application with ease.</p><h3>Django</h3><blockquote><a href="https://www.djangoproject.com/">Django</a> is a high-level Python web framework that encourages rapid development and clean, pragmatic design. — Django</blockquote><p>With Django, you can quickly build an admin panel to manage your application. Django is also backed by a strong Python community that offers many libraries, which are invaluable for quickly building a RAG application.</p><h4>Django Admin</h4><p>Using Django’s admin feature, you don’t need to build authentication, manage permissions, or create a UI to manage your application’s data (sessions, chats, documents).</p><h4>Django REST Framework</h4><p>Django REST Framework allows you to easily implement authentication for your APIs and create other endpoints that will be consumed by the React client to interact with the LLM on the server side.</p><h4>Django Q</h4><blockquote><a href="https://django-q.readthedocs.io/en/latest/">Django Q</a> is a native Django task queue, scheduler and worker application using Python multiprocessing.</blockquote><p>Document embedding and storing embeddings in a vector database can take a significant amount of time. By using this library, you can implement the embedding and indexing process as a task queue, which will be processed separately from your main Django application. Time-consuming tasks will be handled by workers.</p><p>Alternatively, you can use a scheduler to periodically execute tasks to fetch documents or perform other actions. Based on my experience, the UI interface provided by Django Q is very convenient, allowing you to monitor and manage task execution without altering your code. You can even seamlessly retry failed tasks directly from the Django Admin interface.</p><h3>LangChain</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/643/1*_Hp0N19LPcBmZaF4utrq9A.png" /><figcaption>LangChain Environment</figcaption></figure><blockquote>LangChain is a framework for developing applications powered by large language models (LLMs).</blockquote><p>By using <a href="https://python.langchain.com/v0.2/docs/introduction/">LangChain</a>, we can seamlessly build our LLM application from development to deployment. It offers many useful built-in libraries that can accelerate your development process, allowing you to avoid building everything from scratch.</p><h3>ElasticSearch</h3><p>There are many options for vector database technologies available, but personally, I prefer Elasticsearch because I am already familiar with it. In my opinion, it also makes additional filtering of documents easier.</p><h4>Elastron</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8KcsqfY4HL8uu0pRv54Msg.png" /></figure><p>With the help of a tool called <a href="https://elastron.eney.solutions/">Elastron</a>, I can easily manage and debug each index.</p><h3>Implementation</h3><p>I will not cover the entire implementation; instead, I will focus only on the key parts of the application.</p><h4>Document Indexing Triggers</h4><p>We can add document indexing touchpoints or triggers in various parts of the Django Admin interface:</p><p><strong>When the user save document.</strong></p><pre>from django.db import models<br>from django.db.models.signals import post_save<br>from django.dispatch import receiver<br><br>class Document(models.Model):<br>    title = models.CharField(max_length=255, blank=False)<br>    description = models.CharField(max_length=255, blank=True, null=True)<br>    pdf_file = models.FileField(upload_to=&#39;documents/&#39;, blank=False)<br><br>    def __str__(self):<br>        return self.title<br><br><br>@receiver(post_save, sender=TransactionDetail, dispatch_uid=&quot;index_document_to_vector_db&quot;)<br>def index_document_to_vector_db(sender, instance, **kwargs):<br>    # dispatch indexing task to Django Q</pre><p><strong>User can trigger manually using Django Admin bulk actions.</strong></p><pre>from django.contrib import admin<br>from .models import Document<br><br>@admin.action(description=&quot;Index Document to Vector DB&quot;)<br>def index_document_to_vector_db(<br>    modeladmin,<br>    request,<br>    queryset<br>):<br>    # dispatch indexing task to Django Q<br><br><br>@admin.register(Document)<br>class DocumentAdmin(admin.ModelAdmin):<br>    list_display = (<br>        &#39;title&#39;,<br>        &#39;description&#39;,<br>    )<br><br>    actions = [index_document_to_vector_db]</pre><h4>Document Indexing Process</h4><p>Before indexing the document, I recommend converting its content into markdown. There is a tool specifically designed to convert PDF documents into markdown for LLM: <a href="https://pymupdf.readthedocs.io/en/latest/pymupdf4llm/">PyMuPDF4LLM</a>.</p><pre>from langchain_elasticsearch import ElasticsearchStore<br>from langchain_core.documents import Document<br>from .models import Document as DocumentModel<br>from langchain_ollama import OllamaEmbeddings<br>from pymupdf4llm import LlamaMarkdownReader<br><br>ollama_embeddings = OllamaEmbeddings(<br>    model=&quot;llama3.1&quot;,<br>)<br><br>elastic_vector_search = ElasticsearchStore(<br>    es_url=&quot;http://localhost:9200&quot;,<br>    index_name=&quot;&lt;your index name&gt;&quot;,<br>    embedding=ollama_embeddings,<br>    es_user=&quot;elastic&quot;,<br>    es_password=&quot;changeme&quot;,<br>)<br><br>def sync_documents_to_vector_db(documents: list[DocumentModel]):<br>    for document in documents:<br>        md_read = LlamaMarkdownReader()<br>        chunks = md_read.load_data(f&quot;{settings.BASE_DIR}/media/{str(document.pdf_file)}&quot;)<br><br>        for i in range(len(chunks)):<br>            chunk = chunks[i]<br><br>            existing_page = Page.objects.filter(<br>                document__id__exact=document.pk,<br>                page_number=chunk.metadata[&#39;page&#39;]<br>            ).first()<br><br>            if existing_page is None:<br>                Page.objects.create(<br>                    document=document,<br>                    page_number=chunk.metadata[&#39;page&#39;],<br>                    content=chunk.text<br>                )<br>        <br>            elastic_vector_search.add_documents(documents=[<br>                Document(<br>                    page_content=summarization_and_keywords.content,<br>                    metadata={<br>                        **chunk.metadata,<br>                        &quot;document_id&quot;: document.pk,<br>                        &quot;title&quot;: document.title,<br>                        &quot;description&quot;: document.description,                <br>                    },<br>                )<br>            ], ids=[<br>                f&quot;{document.pk}-{chunk.metadata[&#39;page&#39;]}-{i}&quot;<br>            ])</pre><p>You can choose any preferred embedding model you like; in this example, I’m using the Ollama embedding.</p><h4>Document Retrieval from Vector Database</h4><pre>from langchain_elasticsearch import ElasticsearchStore<br>from langchain_core.documents import Document<br><br>ollama_embeddings = OllamaEmbeddings(<br>    model=&quot;llama3.1&quot;,<br>)<br><br>elastic_vector_search = ElasticsearchStore(<br>    es_url=&quot;http://localhost:9200&quot;,<br>    index_name=&quot;&lt;your index name&gt;&quot;,<br>    embedding=ollama_embeddings,<br>    es_user=&quot;elastic&quot;,<br>    es_password=&quot;changeme&quot;,<br>)<br><br>def get_relevant_documents(query: str)<br>    docs: list[Document] = elastic_vector_search.similarity_search(<br>        query=query,<br>        k=50<br>    )<br><br>    return docs</pre><h4><strong>Query to LLM With Retrieved Documents as Context</strong></h4><pre>from langchain_core.prompts import ChatPromptTemplate<br><br>def format_docs(docs) -&gt; str:<br>    return &quot;\n&quot;.join([f&quot;&quot;&quot;=========[1m Document Page Separator [0m=========<br>--- Document Page Metadata ---<br>Document ID: {doc.metadata[&quot;document_id&quot;]}<br>Document Title: {doc.metadata[&quot;title&quot;]}<br>description: {doc.metadata[&quot;description&quot;]}<br>Page: {doc.metadata[&quot;page&quot;]}<br>Document Total Pages: {doc.metadata[&quot;total_pages&quot;]}<br>--- End of Document Page Metadata ---<br>--- Document Content ---<br>{doc.page_content}<br>--- End of Document Content ---<br>=========[1m End of Document Page Separator [0m=========<br>&quot;&quot;&quot; for doc in docs])<br><br>chat_template = ChatPromptTemplate.from_messages(<br>    [<br>        (&quot;system&quot;, &quot;You are an AI Assistant that help human to answer their query based on this given relevant documents:\n{documents}&quot;),<br>        (&quot;human&quot;, &quot;{query}&quot;),<br>    ]<br>)<br><br>docs = get_relevant_documents(query)<br><br>prompt = chat_template.format(<br>    documents=format_docs(docs)<br>    query=query<br>)<br><br>answer = llm.invoke(prompt)</pre><h3>Enhancements</h3><p>You can enhance the implementation in several ways:</p><ul><li><strong>Indexing</strong>: Consider indexing only the summarized version of each page in the vector database and retrieving the full content later to compare the results.</li><li><a href="https://python.langchain.com/v0.1/docs/modules/tools/"><strong>Tools</strong></a>: Explore the concept of tools to give the LLM model additional capabilities that could enhance its responses to queries.</li><li><a href="https://python.langchain.com/v0.1/docs/modules/agents/"><strong>Agents</strong></a>: Utilize LangChain’s agents, where a language model acts as a reasoning engine to determine the appropriate actions to take and the order in which to take them.</li><li><a href="https://python.langchain.com/v0.1/docs/modules/chains/"><strong>Chains</strong></a>: Implement the concept of chains, which refers to sequences of calls — whether to an LLM, a tool, or a data preprocessing step.</li></ul><h3>Conclusion</h3><p>I hope this article has provided you with valuable insights to help you build your first simple RAG application. With the right tools and approach, you can create an effective solution for leveraging LLM technology. Good luck on your development journey!</p><p>If you found this article helpful, please don’t forget to give some claps (you can clap more than once!), leave your comments, and share this article with your friends! If you want to read more personal blog posts, please subscribe to my <a href="https://ghackdev.substack.com/">Substack</a> account! Thank you!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=efcae57755de" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How I Changed My Learning Game in the Age of AI]]></title>
            <link>https://medium.com/@galihlprakoso/how-i-changed-my-learning-game-in-the-age-of-ai-40b0341b7087?source=rss-98d80191145d------2</link>
            <guid isPermaLink="false">https://medium.com/p/40b0341b7087</guid>
            <category><![CDATA[information-technology]]></category>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[learning]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Galih Laras Prakoso]]></dc:creator>
            <pubDate>Sat, 14 Sep 2024 06:32:07 GMT</pubDate>
            <atom:updated>2024-09-14T06:32:07.384Z</atom:updated>
            <content:encoded><![CDATA[<h4>Changing How I Learn in a World with AI. Explore the new strategies I’ve adopted to keep my learning effective and relevant in today’s rapidly evolving landscape.</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yLbIlTbt1R2r0kWv_Hon-A.png" /></figure><p>Hello Hackers! How are you? It’s been quite a while since I last posted a blog. Over the past two weeks, I’ve been very busy, so I didn’t have any spare time to write. But yesterday, I realized I had something interesting to share — it’s about “learning.”</p><p>I’ve been changing my learning methods over time, from practical-first learning to conceptual-first learning and other approaches. However, I still find these methods inefficient because my learning time is getting shorter and shorter.</p><h3><strong>Why do I care so much about improving my learning method?</strong></h3><p>I realized I’m not a fast learner. Some of my close friends say I pick things up quickly, from music to software engineering, but I can’t agree with that. I think it’s because they don’t see the process — they don’t see how hard I work to learn things, or how sometimes I literally hit my head in frustration while learning. When I shared these struggles, I noticed that most people I’ve met experience the same thing. Learning is hard. If you’re a naturally fast learner, consider it a gift — good for you!</p><p>Another reason I care so much about improving my learning method is because AI is becoming “smarter” in important areas. It’s surpassing the intelligence of ordinary humans like me.</p><p>Just yesterday, I read news about how OpenAI improved their LLM model’s performance by adding reasoning capabilities through reinforcement learning. Their latest model can now handle complex reasoning tasks. You can read more about that <a href="https://openai.com/index/learning-to-reason-with-llms/">here</a>.</p><p>The latest OpenAI model has surpassed average human intelligence in several areas. Here are some of its achievements:</p><ul><li><strong>Competitive Programming</strong>: OpenAI’s o1 ranks in the 89th percentile on competitive programming questions (Codeforces) and is among the top 500 students in the US in a qualifier for the USA Math Olympiad (AIME).</li><li><strong>Expert-Level Performance on Scientific Benchmarks</strong>: o1 surpasses human PhD-level accuracy in physics, biology, and chemistry problems (GPQA).</li><li><strong>Mathematical Olympiad Success</strong>: On the 2024 AIME exam, o1 averaged 74% accuracy with a single sample per problem and achieved 93% accuracy using a learned scoring function, placing it among the top 500 students nationally.</li><li><strong>Outperforming Experts on Reasoning Benchmarks</strong>: In reasoning-heavy benchmarks, o1’s performance rivals and even surpasses human experts, with significant improvement over GPT-4o in difficult reasoning tasks.</li><li><strong>Olympiad Programming</strong>: The model competed in the 2024 International Olympiad in Informatics (IOI), ranking in the 49th percentile and scoring as high as 362 points under relaxed competition conditions, surpassing the gold medal threshold.</li><li><strong>Wide-Range Improvement</strong>: o1 outperformed GPT-4o on 54 out of 57 MMLU subcategories and performed competitively with human experts in various reasoning-heavy tasks.</li></ul><h3><strong>My New Way of Learning</strong></h3><p>Learning in this era of technology is significantly different from a few decades ago. From the invention of the first computer, which freed people from complex calculations, to the introduction of the Internet and search engines, people became less reliant on memory because they could simply “Google it.” Now we have ChatGPT — not only can it show search results, but it also helps with complex mental and creative tasks like generating art, photo and video editing, composing music and the most important is that AI now can understands human’s natural language. We can now interact with computers using natural language!</p><p>With these advancements, I believe we need a new way of learning. Simply reading books or following long practical tutorials on YouTube is becoming less effective. We’ll be outperformed by those who use technology in more pragmatic and efficient ways.</p><h4><strong>Focus on Improving What Technology Can’t Do</strong></h4><p>Technology is now more accessible than ever. Everyone has a phone that connects to the internet, making information easy to access. Computers are becoming smarter at understanding natural language, which means they’re also better at solving mental and creative tasks.</p><p>Why read books when you can just ask AI about them? Why improve your English when you can have AI correct your writing? Why bother learning memorization techniques when AI can instantly query a knowledge base to help you remember?</p><p>I think we should focus on areas that AI won’t be able to handle for the next few decades. Here are some skills we can sharpen to ensure we remain relevant in a world where technology continues to advance:</p><h4><strong>Creative Problem-Solving</strong></h4><p>AI is getting better at reasoning and computation, but humans still excel at creativity. Focus on:</p><ul><li><strong>Innovative Thinking</strong>: Combining different fields of knowledge to develop new solutions.</li><li><strong>Complex Problem Framing</strong>: Understanding complex, ambiguous problems that don’t have straightforward solutions.</li></ul><p>For me, I’ve noticed that improving my math skills helps in this area. Developing a deeper conceptual understanding of math has enhanced my thinking. It’s been two months since I decided to study math seriously, and I’ve found it easier to solve practical problems in my daily life.</p><p>I started my math journey by reading the book <a href="https://www.amazon.com/Mathematics-Content-Methods-Meaning-Volumes/dp/0486409163"><em>Mathematics: Its Content, Methods, and Meaning</em></a>. I’m not focusing on competitive math or solving complex equations. Instead, I approach math similarly to how I approach philosophy and religion — through dialectic, contemplation, and meditation on each topic.</p><p>This method has revealed many insights. For example, the book says, “The value of individual numbers comes from their relationships within their respective systems.” I realized this concept correlates with fictional constructs like currency. It’s similar to how we collectively assign value to fictional things as humans.</p><h4><strong>Emotional Intelligence (EQ)</strong></h4><p>AI might be smart, but it lacks the human ability to connect emotionally and socially. Here are key EQ skills to sharpen:</p><ul><li><strong>Empathy</strong>: Understanding and connecting with others on a deep emotional level.</li><li><strong>Interpersonal Communication</strong>: Mastering nuanced communication in personal and professional settings.</li><li><strong>Leadership and Teamwork</strong>: Leading and collaborating effectively, especially in diverse teams.</li></ul><p>In my daily life, I challenge myself to improve my EQ. Getting married has enhanced my emotional connection with my wife, family, and friends. I spend more time socially, connect with new people, and take on leadership roles at work.</p><p>By improving my EQ, I’ve delivered higher impact than ever before. I used to be very introverted, preferring to spend time alone, but now I try to be more social and connected with people around me. I hope this will significantly improve my EQ and help me bring positive impacts to those around me.</p><h4>Adaptability and Lifelong Learning</h4><p>The pace of technological change is accelerating, making the ability to quickly learn new skills vital. Focus on:</p><ul><li><strong>Learning Agility</strong>: Develop a mindset that is open to continuous learning and unlearning.</li><li><strong>Cross-Disciplinary Knowledge</strong>: Be versatile and able to quickly absorb knowledge from various fields, especially fast-evolving ones like biotech, AI, and quantum computing.</li></ul><p>I’m naturally a generalist — a jack of all trades, master of none. I can play some common musical instruments, do graphic design, edit photos and videos, and code complex systems in my professional life. While I’m not exceptionally skilled in any one area, I’ve delivered work using many of these skills.</p><p>I always remind myself not to box myself into one area or become closed-minded. I continuously learn to embrace new things, adapt to new environments, and keep up with technological advancements. Math and emotional intelligence (EQ) have helped me a lot in this area.</p><h4>Critical Thinking and Judgment</h4><p>AI can process vast amounts of data, but humans are still better at understanding context and making value judgments. Focus on:</p><ul><li><strong>Ethical Decision-Making</strong>: Understand the implications of AI and technology on society and be prepared to make ethical choices.</li><li><strong>Evaluating AI Outputs</strong>: Develop the ability to critically assess AI-generated results and make judgments about their validity and relevance.</li></ul><p>In this area, I’ve improved myself by cultivating an interest in philosophy. Learning philosophy has significantly increased my ability to question things — even the most fundamental beliefs.</p><p>Before my journey into philosophy, I was often afraid to question foundational beliefs. But after learning philosophy and questioning many aspects of life, I’ve found that my intellectual foundation has grown stronger. It’s no longer a foundation that limits me, but one that allows me to learn and adapt to new ideas more freely.</p><h4>Complex Interpersonal Skills</h4><p>As AI automates more jobs, human-to-human interaction will become increasingly important. This includes:</p><ul><li><strong>Negotiation</strong>: Navigating complex human dynamics and resolving conflicts.</li><li><strong>Influence and Persuasion</strong>: The ability to inspire and convince others to act in ways that align with shared goals.</li></ul><p>By putting myself in leadership positions, I’ve forced myself to improve my interpersonal skills. When I was a “pure” engineer, everything seemed easier — my only task was to translate business logic into computer code, and the computer would do exactly what I asked. But working with humans is different. Humans are complex. And ultimately, everything we do revolves around humans. By improving interpersonal skills, we can increase our impact and scale it horizontally. Over the next few decades, interpersonal skills will likely remain one of the most valuable abilities — something that AI may struggle to master because we, as humans, understand each other in ways AI cannot.</p><h4>Ethics, Morality, and Philosophy</h4><p>With AI influencing decision-making and societal trends, understanding the ethical implications of technology is crucial. Focus on:</p><ul><li><strong>AI Ethics</strong>: Understand the ethical implications of AI, data privacy, and human-AI interaction.</li><li><strong>Philosophy of Technology</strong>: Explore questions about what it means to live in an AI-dominated world and humanity’s role in it.</li></ul><p>Technological advancements will undoubtedly bring ethical and moral challenges. I remember watching one of Harvard’s open courses on moral philosophy, which sparked deep reflection on these issues.</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FkBdfcR-8hEY%3Flist%3DPLj507hxRXlTFhGVSWc_QgUXifFtU0QVQY&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DkBdfcR-8hEY&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FkBdfcR-8hEY%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/e247311a57fcbae141e633fc6a905fc3/href">https://medium.com/media/e247311a57fcbae141e633fc6a905fc3/href</a></iframe><p>It made me realize that the simple and famous thought experiment called “<a href="https://en.wikipedia.org/wiki/Trolley_problem">The Trolley Problem</a>” has now become very important. With the introduction of self-driving cars controlled by AI, when the AI is driving a car and encounters a situation where it has to choose between a child and an old man, the AI must be able to make this difficult decision in alignment with human ethics. Otherwise, it would be considered dangerous, and of course, we wouldn’t be able to use it at all.</p><p>You can start exploring this concept in a fun way by watching a series on Netflix:</p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FDtRhrfhP5b4%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DDtRhrfhP5b4&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FDtRhrfhP5b4%2Fhqdefault.jpg&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;type=text%2Fhtml&amp;schema=youtube" width="854" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/72b819f49989099ad408811d0575cbd2/href">https://medium.com/media/72b819f49989099ad408811d0575cbd2/href</a></iframe><h4>Human-AI Collaboration</h4><p>The future will likely involve humans working closely with AI systems. Focus on:</p><ul><li><strong>AI Literacy</strong>: Understand how AI works, its limitations, and how to interact with it effectively.</li><li><strong>Hybrid Collaboration</strong>: Learn how to complement AI’s capabilities with uniquely human skills, such as guiding AI in creative tasks or overseeing its use in mission-critical situations.</li></ul><p>Simply learning how AI works conceptually will help us understand how to use it properly. This skill will, of course, become very valuable in the next few decades. Today, everyone has access to AI, but not everyone understands how to use it effectively.</p><h4>Entrepreneurship and Innovation</h4><p>As technology lowers the barriers to entry for new industries, the ability to identify and exploit opportunities will be highly valuable. Key areas include:</p><ul><li><strong>Creating Value with AI</strong>: Innovating by leveraging AI to solve problems or create new markets.</li><li><strong>Business Acumen</strong>: Combining technical knowledge with business skills to bring ideas to market.</li></ul><p>This area is crucial — most of our activities and work are supported by finance. Without the ability to bring innovation to the market properly, there is no sustainability. In this context, I found it easy to improve my skills by building products. For example, I created <a href="https://github.com/galihlprakoso/logseq-plugin-assistseq-ai-assistant">AssistSeq</a>, an AI assistant integrated with my note-taking tool (LogSeq) to help me in my learning journey, leveraging my personal knowledge database. And also building tools that solve specific organizational problem that people are willing to pay for that.</p><p>By building real products, I’ve been able to increase my innovation and entrepreneurial skills. Learning how to add value to my products to solve real-world problems forces me to continuously improve both my entrepreneurial and innovation abilities.</p><h4>Resilience and Well-being</h4><p>The rapid pace of change can be overwhelming. Developing personal resilience will be critical. Focus on:</p><ul><li><strong>Mental Health Management</strong>: Building strategies to manage stress, uncertainty, and the emotional toll of fast-paced change.</li><li><strong>Physical Well-being</strong>: Prioritizing health as a core component of success and adaptability.</li></ul><p>It’s very easy to become stressed nowadays. Everything is growing fast, becoming obsolete just as quickly, and people can become famous or be “canceled” in a single day. Black Mirror episodes are becoming reality, one by one.</p><p>Over the last two years, I also started my journey to maintain my mental health. Learning about Stoicism, Buddhism, and other spiritual wisdom has helped me handle many mental health issues. Of course, I also consult and learn about psychology with my psychologist by scheduling regular sessions to keep my mental health in good condition and enhance my understanding in human psychology.</p><h3>Utilizing Technology</h3><p>Our brain has limitations. We can’t expect it to understand everything quickly or retain all the knowledge we’ve learned instantly. Therefore, we need to leverage available technology to potentially increase our learning speed and quality.</p><h4>Separating Reasoning Skills and Knowledge Base</h4><p>Imagine our brain works like a CPU and RAM, processing inputs and producing outputs. As humans, we also have external ways to store our knowledge. The most conventional method is by taking notes.</p><h4>Reasoning Skills</h4><p>Reasoning skills are developed by practicing critical thinking and constantly validating and challenging our ideas. Methods like the <a href="https://en.wikipedia.org/wiki/Socratic_method">Socratic method</a> help sharpen our opinions and arguments. Becoming more familiar with logical reasoning, math, and philosophy can further enhance our reasoning skills. Reasoning is about connecting the dots — linking knowledge to craft high-quality arguments.</p><h4>Debates and Dialectics with AI</h4><p>I used to engage in deep dialectical discussions with my friends about my ideas, beliefs, and arguments. However, friends have limited time and can’t be available 24/7.</p><p>The solution? Engaging in dialectics with AI. Surprisingly, AI can serve as a sparring partner to debate and test our ideas. AI with reasoning / “slow-thinking” capabilities has been proven to significantly enhance it’s performance, so do with humans. By practicing dialectics and debates with AI, we can dramatically improve our reasoning skills.</p><h4>Second Brain</h4><p>Just like computers, we can’t store all detailed information in our brain, much like a CPU or RAM can’t retain persistent data. Computers store detailed information in persistent storage (HDD, SSD). Similarly, humans store detailed knowledge in notes. However, physical notes or basic note apps can be hard to organize, and information often becomes scattered and difficult to search.</p><p>We need to use technology specifically designed for this purpose. Tools like <strong>Obsidian</strong> or <strong>LogSeq</strong> can help.</p><p>By using <strong>LogSeq</strong>, I’ve enhanced my learning experience. For instance, reading and highlighting books through <strong>LogSeq</strong> has become an essential part of my learning</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xd0PSdsvrT7tiTSB4Qj4qQ.png" /><figcaption>Reading and highlighting book via LogSeq.</figcaption></figure><p>I’ve even built an AI assistant to assist me on this journey, which has proven to be incredibly useful. It can generate random quizzes based on my learning subjects and serve as my dialectic partner to dive deep into complex topics that require advanced reasoning.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*g2yq1A9oNr8M-KrQrnf3BQ.png" /></figure><p>You can learn more about this <strong>LogSeq</strong> plugin <a href="https://github.com/galihlprakoso/logseq-plugin-assistseq-ai-assistant">here</a>.</p><h3>Conclusion</h3><p>In this rapidly evolving world, staying relevant and effective requires more than just keeping up with technological advancements; it demands a proactive approach to learning and self-improvement. By focusing on areas where technology may lag, such as creativity, emotional intelligence, and adaptability, we can complement AI’s capabilities and ensure our skills remain indispensable. Embracing these principles will not only enhance our personal growth but also prepare us for a future where human and AI collaboration will become increasingly integral.</p><p>Remember, leveraging technology to support and amplify our learning is crucial. From using AI for reasoning and dialectical discussions to employing advanced tools for organizing and retaining knowledge, these strategies can significantly boost our efficiency and effectiveness. By continuously improving in these areas, we can navigate the complexities of the modern world and thrive alongside technological advancements.</p><p>I hope you found these insights valuable and inspiring. If you enjoyed this post, please feel free to clap (you can clap more than once! 😝) to show your appreciation. Your feedback means a lot to me, so don’t hesitate to leave a comment with your thoughts and experiences. For more content like this, subscribe to my <a href="https://ghackdev.substack.com/">Substack</a> channel to stay updated on future posts. Thank you for reading, and keep hacking!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=40b0341b7087" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Unlock the Power of LogSeq with AssistSeq: Your Smart AI Companion]]></title>
            <link>https://medium.com/@galihlprakoso/unlock-the-power-of-logseq-with-assistseq-your-smart-ai-companion-2455c265c5f2?source=rss-98d80191145d------2</link>
            <guid isPermaLink="false">https://medium.com/p/2455c265c5f2</guid>
            <category><![CDATA[logseq]]></category>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[large-language-models]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Galih Laras Prakoso]]></dc:creator>
            <pubDate>Thu, 29 Aug 2024 23:28:54 GMT</pubDate>
            <atom:updated>2024-08-29T23:28:54.562Z</atom:updated>
            <content:encoded><![CDATA[<h4>My First LogSeq Plugin: AssistSeq — An Elegant AI Assistant</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*waQW_BxC48us4Pc7LFVGhw.png" /></figure><p>As we all know, LLMs (Large Language Models) have become indispensable in our daily lives. They help us with a wide range of tasks, from brainstorming and summarizing to even coding. Personally, I rely on LLMs every day to assist with work, engage in thought-provoking discussions, and even tackle random topics that pop up. They have truly revolutionized how we approach problem-solving and productivity.</p><p>Beyond my use of LLMs, I also rely heavily on <a href="https://logseq.com/">LogSeq</a> as my “second brain.” It’s an incredible tool for organizing information, allowing me to effortlessly search through my notes and become more efficient in solving problems and learning new things. A few weeks ago, I had a simple yet powerful idea: to combine the capabilities of LLMs with LogSeq, creating a smart AI assistant that fully understands the context of my personal information in a structured way. And thus, <a href="https://github.com/galihlprakoso/logseq-plugin-assistseq-ai-assistant"><strong>AssistSeq</strong></a> was born!</p><h3>LogSeq</h3><p><a href="https://logseq.com/">LogSeq</a> is a privacy-first, open-source knowledge management tool designed for lifelong learners. It functions as a powerful outliner, helping users to structure their thoughts, notes, and projects in a way that enhances productivity and learning. LogSeq’s ability to link notes and create a network of knowledge makes it an invaluable tool for anyone looking to manage information more effectively.</p><h3>LLM</h3><p>LLM stands for <strong>Large Language Model</strong>, a type of artificial intelligence trained on vast amounts of text data. LLMs can understand and generate human-like text, making them useful for a wide range of applications, from conversational agents to content generation and beyond. They are the backbone of many modern AI tools, enabling machines to engage in meaningful conversations and perform complex tasks based on natural language inputs.</p><h3>LangChain</h3><p><a href="https://www.langchain.com/"><strong>LangChain</strong></a> is a powerful framework designed to simplify the creation of applications powered by language models. It provides tools for prompt templating, model chaining, and integration with external services, making it easier to build and customize AI-driven applications. LangChain supports a variety of providers and models, allowing developers to seamlessly switch between them and extend their applications with additional functionality.</p><p>With LangChain, building a chatbot-like assistant becomes a breeze. The framework’s flexibility in handling prompts, chaining models to various tools, and parsing outputs makes it an ideal choice for developing sophisticated AI assistants like AssistSeq. Its support for multiple providers and models also simplifies the integration process, enabling me to work with the tools and models that best suit my needs.</p><h3>Supported Providers</h3><p>Currently, the AssistSeq plugin supports several prominent AI providers, including:</p><ul><li><a href="https://openai.com/"><strong>OpenAI</strong></a>: A leading provider of AI models, known for its powerful and versatile GPT series.</li><li><a href="https://gemini.google.com/app"><strong>Gemini</strong></a>: A platform offering cutting-edge AI solutions tailored for various applications.</li><li><a href="https://ollama.com/"><strong>Ollama</strong></a>: A local provider option, perfect for users who need to work with confidential documents within their LogSeq environment.</li><li><a href="https://groq.com/"><strong>Groq</strong></a>: A provider focused on delivering high-performance AI capabilities.</li></ul><p>These integrations allow AssistSeq to be both flexible and secure, offering users the choice to leverage powerful cloud-based AI or maintain privacy by using local models like Ollama.</p><h3>Key Features</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2Ogt1jW2i3ncWWRuwurSjQ.png" /><figcaption>Contextual Conversation and Smart Visualization</figcaption></figure><h4>Contextual Conversation</h4><p>Unlock deeper insights with <strong>AssistSeq</strong>. This feature automatically reads your current document and indexes all related notes, enabling the AI to engage in meaningful conversations based on comprehensive context. Whether you’re brainstorming ideas or seeking solutions to complex problems, AssistSeq provides tailored responses that are highly relevant to your current work. This contextual awareness transforms AssistSeq into a true assistant, capable of understanding your needs at a deeper level.</p><h4>Smart Visualization</h4><p>AssistSeq transforms complexity into clarity with its <strong>Smart Visualization</strong> feature, powered by <a href="https://kroki.io/">Kroki</a>. Kroki provides a unified API with support for a wide array of visualization tools, including BlockDiag, BPMN, GraphViz, Mermaid, and more. Whether you’re summarizing concepts, explaining ideas, or visualizing data, AssistSeq brings your documents to life with dynamic visual aids. These visualizations help you grasp intricate concepts quickly, making it easier to turn abstract ideas into actionable insights.</p><h4>Search Engine</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6xN8iEIipUllOVk-J59u_A.png" /><figcaption>AssistSeq surfed the internet using Tavily</figcaption></figure><p>AssistSeq’s <strong>Search Engine</strong> feature, powered by <a href="https://tavily.com/">Tavily</a>, allows the AI to surf the internet and fetch information that isn’t readily available in your current documents. Tavily is a search engine optimized for LLMs and RAG (Retrieval-Augmented Generation), designed to deliver efficient, quick, and persistent search results. With this feature, AssistSeq can pull in external knowledge to enhance the context of your discussions, making it a powerful tool for research and information gathering.</p><h4>Web Scrapping</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*f_CNPCJ-glvQWZV1BCIDNg.png" /><figcaption>AssistSeq scrapped the website using Cheerio</figcaption></figure><p>Whenever AssistSeq encounters a URL, it automatically scrapes the website for relevant information and incorporates it into the conversation context. This <strong>Web Scraping</strong> feature is powered by Cheerio, a fast and flexible web scraping tool. Cheerio enables AssistSeq to extract data from web pages in a manner similar to jQuery, making it easy to integrate fresh content from the web directly into your LogSeq workspace. This ensures that your AI assistant always has the most up-to-date information at its disposal.</p><h3>Conclusion</h3><p>Creating <strong>AssistSeq</strong> has been a rewarding experience, blending the power of LogSeq with the intelligence of modern LLMs to create a truly smart assistant. I invite you to try out <a href="https://github.com/galihlprakoso/logseq-plugin-assistseq-ai-assistant">AssistSeq</a> and elevate your LogSeq experience with this smart AI tool!. If you enjoyed this post, be sure to follow my Medium channel and subscribe to my <a href="https://ghackdev.substack.com/">Substack</a> for more personal insights and awesome blog posts. Let’s continue exploring the limitless possibilities of AI together!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2455c265c5f2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Hyper-Efficient or Hyper-Inefficient? The Dangerous Mindset of Stagnant Engineers]]></title>
            <link>https://medium.com/@galihlprakoso/hyper-efficient-or-hyper-inefficient-the-dangerous-mindset-of-stagnant-engineers-999b88dcdc02?source=rss-98d80191145d------2</link>
            <guid isPermaLink="false">https://medium.com/p/999b88dcdc02</guid>
            <category><![CDATA[productivity]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[tech]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Galih Laras Prakoso]]></dc:creator>
            <pubDate>Tue, 27 Aug 2024 23:59:15 GMT</pubDate>
            <atom:updated>2024-08-28T02:41:03.786Z</atom:updated>
            <content:encoded><![CDATA[<h4>How Attitude Impacts the Quality of Work as a Software Engineer: Why It’s Critical, and Perhaps the Most Important Factor</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CgWqQSvVRDkhx0GhOTAESg.png" /><figcaption>Generated by DALL·E</figcaption></figure><p>Let me start with my own experience. After five years in the industry, I’ve interviewed and worked with numerous software engineers. I’ve encountered many great engineers whose quality of work and the impact they deliver to products and businesses are truly outstanding. However, I’ve also met engineers — some of them quite senior, with over five years of experience — who seem “stagnant,” lacking both soft and hard skills.</p><h4>What causes them to become “stagnant”?</h4><p>When I tried to identify a pattern among these stagnant engineers, I had a realization: it all comes down to <strong>attitude</strong>. Attitude shapes the quality of a software engineer over the long term. As John Maxwell discusses in his book <em>Attitude 101</em>, a person’s attitude can significantly influence their growth or stagnation, ultimately determining whether they excel or falter in their career.</p><blockquote>Your attitude is like the minds paintbrush. It can paint everything in bright, vibrant colors-creating a masterspiece. — John C. Maxwell</blockquote><blockquote>Attitude, to me, is more important than education, than money, than circumstances, than failures, than successes, than what other people think or say or do. It is more important than appearance, giftedness or skill. It will make or break a company… a church… a home. — John C. Maxwell</blockquote><h3>The Dangers of Bad Attitudes</h3><p>Many software engineers, especially as they gain more experience, fall into the trap of arrogance. They become too self-assured, believing that they no longer need to learn from anyone who isn’t as senior as they are. This arrogance limits their ability to stay humble, to descend from the mountaintop, and to embrace their failures and mistakes. Over time, these attitudes make them dull, insensitive, and unable to adapt.</p><p>Bad attitudes can lead to a decline in a software engineer’s quality of work. They start to believe they know everything, dismissing input from others and refusing to accept that they might be wrong. This mindset not only stifles their growth but also damages their relationships with peers and colleagues.</p><h3>Hyper-Efficiency vs. Hyper-Inefficiency</h3><p>I’ve noticed that stagnant senior engineers because of they have “bad” attitudes often fall into two categories: hyper-efficient and hyper-inefficient.</p><h4>The Hyper-Efficient Engineer</h4><p>Hyper-efficient engineers adhere strictly to certain theories and are fiercely opinionated. They impose their rigid beliefs on others without considering the context or the current situation. They do what they believe is best, closing their ears and eyes to other perspectives, and holding onto their dogmatic views at all costs.</p><p>I once worked on a project as a frontend engineer where a small data-sorting issue arose. The backend didn’t provide the necessary parameter to change the sorting of my data. Requesting this change would have taken weeks or even longer, so I decided to tackle the backend issue myself. I expected to find a simple raw SQL query or Database ORM where I could add ASC or DESC to handle the sorting. But after an hour of digging, I discovered something surprising: the sorting logic was in the application layer. The engineer responsible had chosen to sort the data there rather than in the database. When I asked him about it, he responded, “It’s easier to scale the application than to scale the database.” I was taken aback. We only had 10 users, all of whom were engineers, product managers, or executives—why was he already worrying about scalability for an admin panel that might eventually have less than 20 users?</p><p>Engineers like this waste time by focusing on unnecessary details. They are overly opinionated and can drive their peers, especially juniors, crazy. Their stubbornness harms the people around them, the product, and the business. It also stunts their own growth, making them unable to adapt and evolve, leading to stagnation.</p><h4>The Hyper-Inefficient Engineer</h4><p>On the other hand, there are senior engineers who are incredibly inefficient. They work long hours not because they overthink, but because they never address their technical debt, which compounds and slows them down. Or perhaps they are simply lazy, yet fortunate enough to remain employed.</p><p>These engineers lack dignity, ownership, principles, and the commitment to deliver high-quality work. Their standard of ownership barely reaches the “as long as it works” bar. Over time, this kind of attitude cripples the team with massive technical debt, forcing others to pay for it.</p><p>I once encountered a senior engineer whose code was a mess — hardcoded values everywhere, even secret keys. His work was the antithesis of “Clean Code.” Instead of following standard practices, he invented bizarre, convoluted solutions that made his work difficult to understand. He was opinionated, but his opinions were “weird” and far removed from industry standards.</p><p>He did no testing and made no effort to improve the quality of his work. When I first saw his code, I thought the messiness might be due to the pressure of building an early-stage product. But no, it was peacetime, and he had plenty of spare time. His recent commits were no different from those made years ago. His poor system architecture and pervasive code smells showed that he had no intention of addressing his technical debt. Engineers like him cripple the team, worsening the situation when the product and business need to deliver quickly to stay competitive or meet strategic goals.</p><h3>Conclusion</h3><p>Attitude is crucial to the quality of work as a software engineer. It shapes who we become and how we perform over the long term. To avoid becoming a “hyper-” engineer, it’s important to remain humble, open to learning, and adaptable. Don’t let arrogance cloud your judgment or stifle your growth. Pay attention to your technical debt and always strive to improve your work. By cultivating a positive attitude, you’ll continue to grow, adapt, and deliver high-quality work throughout your career.</p><h3>Closing Thoughts</h3><p>I’d love to hear your thoughts and experiences on this topic. Feel free to share them in the comments section below. If you enjoyed this post, follow my Medium channel, and don’t forget to subscribe to my <a href="https://ghackdev.substack.com/">Substack</a> for more personal and interesting blog posts.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=999b88dcdc02" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Fintech’s Multi-Platform Dilemma: Flutter or React Native?]]></title>
            <link>https://medium.com/@galihlprakoso/fintechs-multi-platform-dilemma-flutter-or-react-native-1a14d221b9f4?source=rss-98d80191145d------2</link>
            <guid isPermaLink="false">https://medium.com/p/1a14d221b9f4</guid>
            <category><![CDATA[flutter]]></category>
            <category><![CDATA[react-native]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[front-end-development]]></category>
            <category><![CDATA[mobile-app-development]]></category>
            <dc:creator><![CDATA[Galih Laras Prakoso]]></dc:creator>
            <pubDate>Sat, 10 Aug 2024 20:12:24 GMT</pubDate>
            <atom:updated>2024-08-10T20:12:24.506Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qoBi_Y0pTXp-UQyuG4QBuA.png" /><figcaption>Generated by DALL·E — ChatGPT</figcaption></figure><p>Currently, I work for a digital payment company in Indonesia called <a href="https://yourpay.co/">YourPay</a>, a mobile banking service that helps Indonesian Migrant Workers and their families conduct transactions overseas.</p><p>YourPay is currently available on a single platform, Android, with the app written using the Native Android framework. Recently, the company decided to expand to iOS. Instead of maintaining separate native frameworks for iOS and Android with two distinct teams, the company chose to move towards a multi-platform/hybrid approach.</p><h3>Flutter vs React Native</h3><p>Switching to a multi-platform framework was a significant decision. We needed to ensure that we could deliver our product quickly and efficiently while considering key factors like performance, security, and scalability. Among the available frameworks, React Native and Flutter are the top contenders in the multi-platform application space.</p><p>Deciding which framework to use was a challenging task for me as a Frontend Lead, especially during my first week at the company. I had to weigh many factors before arriving at the “best” decision, considering both technical and business aspects.</p><h4>Security</h4><p>Security was our top priority since our company relies heavily on user trust. Ensuring that our users are fully protected is critical. In this regard, Flutter stood out as the better option.</p><p><strong>React Native</strong></p><p>React Native consists of two main components: JavaScript and Native. The application’s business logic resides on the JavaScript side, within a file known as main.jsbundle, which can be &quot;easily&quot; reverse-engineered.</p><p>This vulnerability could allow someone to manipulate the app&#39;s logic, potentially leading to malicious behavior, especially in an app that handles large transactions for users who may not be tech-savvy. You can see how <a href="https://vlourme.medium.com/reverse-engineering-react-native-and-hermes-byte-code-bb5b96db368f"><strong>this guy</strong></a> did, he could manipulate the javascript part to alter the logic and unlocking unexpected behaviour of the app.</p><p>Even with the introduction of Hermes — a JavaScript engine optimized for React Native — JavaScript can still be decompiled using tools that are readily available, posing a security risk. You can see the detailed step how to do it in <a href="https://labs.cognisys.group/posts/How-to-Decompile-Hermes-React-Native-Binary/"><strong>this article</strong></a>.</p><p><strong>Flutter</strong></p><p>Flutter, on the other hand, is “safer” because it compiles directly to native code. While no system is entirely secure, Flutter’s approach makes it more challenging to reverse-engineer. Unlike React Native, which relies on a bridge to communicate between JavaScript and native code, Flutter compiles directly to native code, reducing the risk of tampering.</p><p>While Flutter isn’t entirely immune to reverse engineering, the process is more complex and constrained compared to React Native. You can take a look to these two tools that built specifically to help reverse engineers to do their work:</p><ul><li><a href="https://github.com/worawit/blutter">https://github.com/worawit/blutter</a></li><li><a href="https://github.com/ptswarm/reFlutter">https://github.com/ptswarm/reFlutter</a></li></ul><p>Despite Flutter’s advantages in security, it doesn’t make Flutter invulnerable to attacks. Similarly, React Native can be made more secure through various methods. However, in this context, I prefer Flutter due to its stronger security features.</p><h4>Accessability</h4><p>Accessibility was our second most important concern. Some of our users are not tech-savvy and use low-end devices, starting from Android 5.1.1. In this area, Flutter was the clear winner.</p><p><strong>React Native</strong></p><p>The latest React Native version requires a minimum Android SDK version of 23 (Android 6.0). Choosing React Native would mean leaving behind users still on Android 5.x.x, which isn’t an option given that many of them still rely on our app for essential transactions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*x1654OTe1rGPZWZRnLoTCQ.png" /></figure><p><strong>Flutter</strong></p><p>Flutter, however, still supports Android Lollipop (API 21–5.x.x), ensuring that all of our customers can continue to access and use our app without any issues.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lNSVmcm0za4U6aw7mZnRBA.png" /></figure><h4>Performance</h4><p>Performance is a critical factor for us, as we prioritize functionality over aesthetics or other non-essential features. Our goal is to create a simple, user-friendly app that meets our customers&#39; needs efficiently. In terms of performance, Flutter emerged as the winner.</p><p><strong>React Native</strong></p><p>React Native is having an architecture where there are two main sides: Javascript side and Native side and there is a bridge connecting those two sides.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/850/1*gK4Fc81djEbTzhzsavOI7w.png" /><figcaption><a href="https://www.researchgate.net/figure/Overview-of-the-React-Native-architecture_fig8_355670705">https://www.researchgate.net/figure/Overview-of-the-React-Native-architecture_fig8_355670705</a></figcaption></figure><p>React Native uses bridge to connect Javascript that contains most of the business logics with native modules. The JavaScript code runs in a separate thread and communicates with the native side, which can introduce latency, especially in complex applications with heavy UI interactions.</p><p>React Native can also achieve smooth animations, but it may struggle with performance in more complex scenarios due to the overhead of the JavaScript bridge. This can result in dropped frames and a less smooth user experience.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IWTlm5HnW7Asx_bX0vBs1Q.png" /><figcaption><a href="https://reactnative.dev/blog/2018/06/14/state-of-react-native-2018#architecture">https://reactnative.dev/blog/2018/06/14/state-of-react-native-2018#architecture</a></figcaption></figure><p>How about the latest improvements on the “bridge”? Yeah, there should be a lot of improvements in the future, one of them included to current latest version right now by introducing JSI and there might be more improvements coming in the next version of React Native.</p><p><strong>Flutter</strong></p><p>Flutter, in contrast, uses the Skia graphics engine to render UI directly to the canvas, bypassing native UI components. This approach allows Flutter to deliver near-native performance, with smooth animations at 60 FPS or even 120 FPS on devices that support higher refresh rates.</p><p>Flutter’s direct rendering approach also reduces overhead, making it the better option in terms of startup time, battery consumption, and memory usage.</p><h4>Team</h4><p>Our existing mobile app team has a strong background in Native Development (Android and iOS), with extensive experience in Java, Kotlin, and Swift — languages that are object-oriented and strongly-typed by default. For this reason, Flutter was the better choice.</p><p><strong>React Native</strong></p><p>React Native’s business logic is written in JavaScript or TypeScript. While it’s possible to write on the Native side using Native Modules, most of the development time would be spent on JavaScript or TypeScript.</p><p>Transitioning to functional programming with a language as flexible as JavaScript was challenging for our team, who found adapting to TypeScript difficult given their extensive background in Kotlin, Java, and Swift.</p><p><strong>Flutter</strong></p><p>Flutter, on the other hand, uses Dart — a language less widely known than JavaScript but similar in syntax to Java, C#, or Swift.</p><p>This similarity made Dart easier for our team to learn, as they were already familiar with object-oriented programming and strong typing. Dart’s structure and focus on object-oriented programming provided a smoother transition for our team.</p><h4>Scalability</h4><p>Our company is currently stable in terms of growth — slow but steady. While we’ve achieved product-market fit, we’re focused on profitability and sustainability rather than “hyper-growth”. However, we have exciting and ambitious projects for the future, and scalability is essential to our long-term vision. In this aspect, Flutter proved to be the better choice.</p><p><strong>React Native</strong></p><p>Having worked on several projects using React and React Native with TypeScript, I’ve found it challenging to keep React code clean as projects grow in complexity. JavaScript’s flexibility, while beneficial in smaller projects, can lead to maintainability challenges in larger applications, even with TypeScript.</p><p><strong>Flutter</strong></p><p>Flutter, with Dart, promotes maintainable and readable code, which is crucial as projects become more complex. Dart’s strong typing and object-oriented features make it well-suited for building scalable applications.</p><p>Although JavaScript is more popular and widely used in web development, I believe learning Dart isn’t difficult, especially for developers familiar with languages like Java, C#, and Swift.</p><h3>Conclusion</h3><p>Choosing the right multi-platform framework is not easy. There’s no “best” framework that fits every situation. It all depends on the specific needs, context, and conditions of your project. I hope this blog post has provided you with insights to make better decisions when choosing a multi-platform framework.</p><p>Don’t forget to clap, share, and drop your comments below! If you’d like more personal and interesting posts, please follow my <a href="https://ghackdev.substack.com/">Substack</a> channel. Thank you!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1a14d221b9f4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why Your Engineering Productivity Metrics Are Failing: Insights from Google’s GSM Framework]]></title>
            <link>https://medium.com/@galihlprakoso/why-your-engineering-metrics-are-failing-insights-from-googles-gsm-framework-805aa586e42c?source=rss-98d80191145d------2</link>
            <guid isPermaLink="false">https://medium.com/p/805aa586e42c</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[software-development]]></category>
            <dc:creator><![CDATA[Galih Laras Prakoso]]></dc:creator>
            <pubDate>Thu, 01 Aug 2024 04:13:36 GMT</pubDate>
            <atom:updated>2024-08-01T11:43:18.165Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LL50AJ1PMzdnRbqS-GhaQw.jpeg" /><figcaption>Generated by Copilot — Microsoft Bing</figcaption></figure><p>Currently, I’m preparing for my first day as an Engineering Lead at my new company. To get ready, I’ve been reading extensively — articles, books, and listening to podcasts about engineering leadership. I’ve also been speaking with friends who have held engineering lead or manager roles for several years. In this blog post, I will share a valuable framework I found in the book “<a href="https://www.amazon.com/Software-Engineering-Google-Lessons-Programming/dp/1492082791">Software Engineering at Google</a>” called GSM (Goals/Signals/Metrics). This framework is instrumental in selecting meaningful metrics with clear goals.</p><h3>Engineering Productivity Metrics</h3><p>In my experience as a Software Engineer, I’ve encountered numerous approaches to measuring engineering productivity. Different metrics serve various purposes and are measured in diverse ways. The most common metrics I’ve used include:</p><ul><li><strong>Number of Merged PRs</strong>: This metric reflects engineering productivity on a micro-level.</li><li><strong>Number of Deployments/Releases</strong>: This slightly higher-level metric represents engineering productivity. Deployments can include new features, bug fixes, code refactoring, and other non-feature-related updates.</li><li><strong>Test Coverage (Backend)</strong>: In my experience this metric is typically measured on the backend, while the frontend focuses on integration testing (e2e).</li><li><strong>Number of Bugs and Resolved Bugs</strong>: This metric ensures code quality and tracks how quickly engineers resolve issues.</li></ul><p>These metrics are useful but should not be the sole approach to measuring productivity. I’ve worked under a leader who blindly implemented productivity measurements without considering other factors, resulting in poor decisions that hurt team trust, motivation, and overall product quality.</p><h3>GSM Framework</h3><p>Instead of relying solely on traditional metrics, we can use Google’s GSM Framework to guide metric creation. The framework involves defining:</p><ul><li><strong>Goal</strong>: A high-level objective without specific implementation details.</li><li><strong>Signal</strong>: Information indicating whether the goal is achieved, which usually cannot be measured directly.</li><li><strong>Metric</strong>: A measurable proxy for signals.</li></ul><p>Following this framework from Goals to Signals to Metrics can prevent many issues. In my past experiences, hastily creating metrics without defining clear goals and signals led to miscommunications, wasted time, and poor decisions. For instance, upper management reviewing reports without context might misinterpret the data, leading to misguided actions.</p><h3>Streetlight Effect</h3><p>A funny anecdote from my hometown illustrates this concept:</p><blockquote>A man was looking for his lost wallet under a streetlight at night. Another man asked him where he dropped it, and the first man replied, “In the bushes at the edge of the forest.” Confused, the second man asked, “Then why are you looking here?” The first man innocently replied, “Because it’s dark in the bushes, and there’s no light there.”</blockquote><p>This situation, known as the “streetlight effect,” highlights the danger of looking only where it’s easy rather than where the answer lies. The GSM framework helps avoid this by ensuring we define goals and signals before metrics.</p><h4>Metrics Creep and Metrics Bias</h4><p>Ciera Jaspen, in “Software Engineering at Google,” warns against metrics creep and metrics bias. When metrics are chosen without a principled approach and fail to meet stakeholders’ expectations, stakeholders might suggest new metrics to achieve desired results. This can harm the engineering team over time. The GSM framework prevents these issues by establishing a clear, goal-oriented approach to metrics.</p><h3>Conclusion</h3><p>The GSM framework offers a structured approach to selecting meaningful metrics in engineering leadership. By defining goals and signals before metrics, we can avoid common pitfalls and ensure our measurements accurately reflect productivity and quality.</p><p>Thank you for reading! If you enjoyed this post, please give it a clap, drop a comment, share it with others, and follow my Medium blog. For more frequent and personal insights, follow my blog on <a href="https://ghackdev.substack.com/"><strong>Substack</strong></a>. Your support means a lot!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=805aa586e42c" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>