<?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 Bagus Aryabima on Medium]]></title>
        <description><![CDATA[Stories by Bagus Aryabima on Medium]]></description>
        <link>https://medium.com/@bagus.rahman.aryabima?source=rss-c591be081b7d------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*9rSTRY0DSYqXTPDCg72qZQ.jpeg</url>
            <title>Stories by Bagus Aryabima on Medium</title>
            <link>https://medium.com/@bagus.rahman.aryabima?source=rss-c591be081b7d------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 21:01:56 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@bagus.rahman.aryabima/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[140th Monthly Technical Session]]></title>
            <link>https://medium.com/henngeblog/140th-monthly-technical-session-fb5fbcfacb60?source=rss-c591be081b7d------2</link>
            <guid isPermaLink="false">https://medium.com/p/fb5fbcfacb60</guid>
            <category><![CDATA[hennge]]></category>
            <category><![CDATA[puredata]]></category>
            <category><![CDATA[monthly-technical-session]]></category>
            <category><![CDATA[llm]]></category>
            <category><![CDATA[genai]]></category>
            <dc:creator><![CDATA[Bagus Aryabima]]></dc:creator>
            <pubDate>Tue, 17 Nov 2026 02:01:06 GMT</pubDate>
            <atom:updated>2026-11-17T02:01:06.023Z</atom:updated>
            <content:encoded><![CDATA[<p>HENNGE’s Monthly Technical Session (MTS) is a platform for members to share knowledge with their peers. Each speaker gives a talk for 10–20 minutes, followed by a Q&amp;A session. Occasionally there is also a lightning talk session at the end of an MTS, in which speakers give 5-minute talks.</p><p>The 140th MTS was held on March 27, 2026.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ooscTQviR1D-iUIimUKVMQ.jpeg" /><figcaption>Rob being the MC of the 140th MTS</figcaption></figure><p><strong>“Good Taste: Recipes to Simplify Your Code” by Ed</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/996/1*Gr2TP0ZlgGLApwJvUJi5LQ.png" /><figcaption>Ed presenting “Good Taste: Recipes to Simplify Your Code”</figcaption></figure><p>Code is read more than it is written. Good taste is about reducing mental load for the reader. Ed shared several recipes to simplify code, beginning with “Become a Never-Nester”. Each level of nesting adds another condition the brain must hold. To de-nest, consider doing inversion (flipping conditions and returning early) or extraction (pulling inner blocks into their own functions).</p><p>Ed’s second recipe is “Simplify Your Conditionals”. If a conditional contains more than one condition, consider extracting it.</p><p>Ed’s third recipe is “Think Twice About Booleans”. Booleans can be problematic because they only have two states (e.g., sometimes a third option is necessary), discards context (e.g., <em>when</em> did something happen?), and new requirements may result in multiple boolean flags that may be mutually exclusive. Instead of storing boolean values, consider deriving booleans from richer data. As opposed to mutually exclusive boolean flags, use enums to enforce valid states.</p><p>Ed’s final recipe is “Lean on Your Language”. For example, replace if/elif chains with data structures.</p><p>All these recipes reduce cyclomatic complexity, i.e., the number of independent paths through your code. Fewer branches reduces complexity, which increases readability, which increases maintainability, which finally makes it easier to add features and/or fix bugs.</p><p><strong>“align-items: despair;” by Jeremi</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/996/1*CEOkj179MT63w8e3viR2rw.png" /><figcaption>Jeremi presenting “align-items: despair;”</figcaption></figure><p>The struggle to center a &lt;div&gt; has been one of the longest running jokes of web development. This struggle actually reflects how the web has transformed from a document-based system into a platform for complex, app-like interfaces.</p><p>Looking back to the days of HTML 1.0 in 1991 allows us to understand that the web was never meant to do layout. Early centering was done with &lt;center&gt;&lt;/center&gt;. This worked visually, but only for horizontal centering. It also unfortunately had poor semantics and limited control.</p><p>The first layout model came by way of HTML 2.0/3.0 and CSS 1, circa 1995–1997. The core concept is that every element is a box. Layout tools such as block, inline, float, and absolute positioning were introduced. Horizontal centering can be done with “margin: 0 auto”, but such approaches worked only when the container had fixed size and there are some free space.</p><p>By the time of HTML 4 and CSS 2 in around 2002–2014, developers relied on hacks due to the growing requirements and insufficiency of native tools. For example, approaches involving “float: left”, which has the original purpose of wrapping text around images, are used for full page layout. This includes, but not limited to, multi-column layouts, grid-like pages, and entire site structure. Another example is a vertical centering trick by making “line-height” equal to the element height, even if it only works reliably for single-line texts. Other developers do vertical centering by faking table behavior and using “vertical-align: middle”, despite this meaning using table semantics for non-table layout.</p><p>HTML 5, specifically around 2012–2016, made flexbox available and widely practical. At last there is a layout model for UI axes. It worked because of its main axis + cross axis model, automatic space distribution, and responsive-first behavior.</p><p>According to Jeremi, the next step is CSS Grid, which introduced two dimensional layout control. It gives us rows and columns as first-class concepts, fewer hacks + clearer intent, and layout systems (instead of accidents).</p><p><strong>“Apocalyptic Idiocy” by Tina</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/996/1*zkktciulXIsg1mPwjfPy7w.png" /><figcaption>Tina presenting “Apocalyptic Idiocy”</figcaption></figure><p>Tina explains the concerns regarding the widespread adoption of Generative AI. For one, Generative AI are energy-intensive. 10 times more electricity is being consumed per query compared to search engines. It is estimated that 1,287 megawatt hours is consumed to train OpenAI’s GPT-3, which is enough to power about 120 homes in USA for 1 year. The predicted electricity usage of AI data centers in general goes as high as 1,050, terawatt hours.</p><p>Infrastructure of Generative AI requires substantial water for cooling. 1 million liters is used for a single training cycle of Gemini. 3.75–9.5 million liters is used to cool server to support ChatGPT’s 2.5 billion prompts daily. The predicted daily water consumption of data centers for 2050 goes as high as 28.11 billion liters.</p><p>Generative AI also affects the human brain. MIT researched the brain of 54 people with similar education background over the span of 4 sessions in 4 months. It was discovered that 83.3% can’t quote the content essay they just generated with LLM, 55% lower brain connectivity than the brain-only group, and 100% of the LLM-only group can’t do correct quoting. Other researches discovered reduced critical thinking with reliance to Generative AI (due to the brain effort shifting from problem solving to verifying AI response), delusional experience in relation to intense chatbot usage, and 16 cases of chatbot-related deaths recorded as of February 2026.</p><p>Tina doesn’t have an exact answer as to what should be done, but she suggests trying to use Generative AI selectively and intentionally or limit its use as much as possible. Generative AI can be useful, but it’s not a magical one size fits all solution. In whatever way, the cost will eventually come.</p><p><strong>“Coding With Confucius: Can We Decouple Logic From the English Language?” by Eren</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/996/1*nzleEOF_pY2WJ2vailO77g.png" /><figcaption>Eren presenting “Coding With Confucius: Can We Decouple Logic From the English Language?”</figcaption></figure><p>Eren explores how <a href="https://wy-lang.org/">wenyan-lang</a> incorporates Chinese philosophy into coding. wenyan-lang is a Turing-complete language written entirely in traditional Chinese characters. It transpiles into modern languages such as JavaScript, Python, Ruby, and C++.</p><p>wenyan-lang’s design is based on the 2 pillars of ancient Chinese philosophy, i.e., Confucianism and Taoism. The former is about social order and structure, while the latter is about flow.</p><p>Confucianist principles can be found in wenyan-lang’s variable declaration. Instead of simply assigning values, there is a poetic three-step process: existence, being, and rectification. Naming is also a very important topic, e.g., “a father has to be a father”. Assigning the wrong variable name is seen as breaking the etiquette of the code and disrupting social order.</p><p>Taoist concepts regarding the nature of the universe is incorporated into wenyan-lang. For example, Instead of binary true and false booleans, the language uses Yin and Yang to represent intertwined states. Another example is that arrays in wenyan-lang start at one rather than zero. This symbolizes the Taoist belief that everything stems from a primordial state and that “nothing” cannot exist once things begin to stem from the source.</p><p>Eren suggests that language design encodes worldview. For instance, most modern languages are imperative and give commands to machines, while wenyan-lang views the machine as part of nature, treating code as literature or prose.</p><p><strong>“The Root of All Evil” by Salma</strong></p><p>Donald Knuth asserts in 1974 that premature optimization is the root of all evil. Salma exemplifies this by presenting a hypothetical system architecture of a web application that goes to the very deep end of premature optimization. In the pursuit of excellence, onboarding a new developer is going to take months and require a PhD in distributed systems, while at the same time bankrupting the company in AWS data transfer fees.</p><p>Premature optimization is the root of all evil because the system is spending most of its CPU cycles managing its own infrastructure (e.g., Kubernetes polling, Kafka heartbeats) rather than serving user traffic. It also sacrifices code readability and maintainability.</p><p>Salma explains the things Donald Knuth suggests to do instead of premature optimization.</p><ul><li>Correctness first: Write code to be clean, readable, correct, and easily maintainable. Ignore micro-efficiencies during the initial draft.</li><li>Measure, don’t guess: Once the code is working, use profiling tools to objectively measure performance and identify the actual bottlenecks, i.e., the “critical 3%”.</li><li>Target the 3%: Apply rigorous, heavy optimization <em>only</em> to the measured actual bottlenecks.</li><li>Leave the 97% alone: Accept that most of the code doesn’t need to be fast, it just needs to be understandable.</li></ul><p><strong>“Pd, Programmable Music!” by Sam</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/996/1*56xWNGAa4PTUsP3D4H4ShA.png" /><figcaption>Sam presenting “Pd, Programmable Music!”</figcaption></figure><p><a href="https://puredata.info/">Pure Data</a> (Pd) is an open-source, real-time, graphical programming language for audio and graphical processing. Pd is versatile and is used for live performances, audio processing (e.g., gaming, embedded programming), electronic instrument design, and interactive installation art. It is often used for more experimental works, as it enables creativity but with the cost of increased complexity.</p><p>libPd is the C-based audio engine that powers Pd. It is designed to allow Pd to be embedded into various platforms. The community has integrated libPd into numerous environments, including macOS, Windows, iOS, Android, and Meta Quest. Sam himself has had some experience with the last one, creating an augmented reality musical instrument using Pd.</p><p>Programs in Pd are referred to as “patches”. Patches consists of objects (e.g., oscillators), messages, numbers, bangs (i.e., button), toggles, vsliders, etc. Sam walked us through some of the patches he made. The “Hello, World!” of Pd is basically a 440 Hz tone. Another patch he made is an FM synthesizer. By far the most complex patch Sam shared with us is his drum machine. It involves a metronome and allows users to choose instruments (e.g., kicks, snares, hats).</p><p>As always, we had a party afterwards 😄</p><p>See you at the next MTS!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fb5fbcfacb60" width="1" height="1" alt=""><hr><p><a href="https://medium.com/henngeblog/140th-monthly-technical-session-fb5fbcfacb60">140th Monthly Technical Session</a> was originally published in <a href="https://medium.com/henngeblog">henngeblog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[137th Monthly Technical Session]]></title>
            <link>https://medium.com/henngeblog/137th-monthly-technical-session-a0b39ce4a698?source=rss-c591be081b7d------2</link>
            <guid isPermaLink="false">https://medium.com/p/a0b39ce4a698</guid>
            <category><![CDATA[beatles]]></category>
            <category><![CDATA[hennge]]></category>
            <category><![CDATA[monthly-technical-session]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[town-hall]]></category>
            <dc:creator><![CDATA[Bagus Aryabima]]></dc:creator>
            <pubDate>Fri, 26 Jun 2026 05:01:03 GMT</pubDate>
            <atom:updated>2026-06-26T05:01:03.011Z</atom:updated>
            <content:encoded><![CDATA[<p>HENNGE’s Monthly Technical Session (MTS) is a platform for members to share knowledge with their peers. Each speaker gives a talk for 10–20 minutes, followed by a Q&amp;A session. Occasionally there is also a lightning talk session at the end of an MTS, in which speakers give 5-minute talks.</p><p>The 134th MTS was held on December 19, 2025.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*opOm4a_X_Ck4mF-lQ6K5Jw.jpeg" /><figcaption>Nektar being the MC of the 137th MTS</figcaption></figure><p><strong>“Accessibility in Frontend: Lessons From My Internship” by Pawel</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*RSXswWRHGrFZTC13N8WNwA.jpeg" /><figcaption>Pawel presenting “Accessibility in Frontend: Lessons From My Internship”</figcaption></figure><p>Pawel used to think accessibility is something that is nice to have, but his experience of being trapped in his own UI modal due to poor keyboard navigation changed his mind.</p><p>What happened was:</p><ul><li>Tab key did not work</li><li>Escape key did not close the modal</li><li>Focus disappeared</li><li>The page behind the modal was still scrollable</li></ul><p>Focus management was the biggest lesson learned from this experience. When a modal opens, focus must move inside it and nowhere else. When the modal closes, focus must return exactly where the user was before. Otherwise, keyboard users will instantly be lost. Speaking of keyboard support, intuitiveness is important. For example, Tab key (as well as Shift + Tab) is for navigation and Escape key is for closing a modal.</p><p>Pawel also learned to use ARIA only when semantics are missing from native HTML. Using ARIA to paper over a broken structure is likely to cause more problems than it fixes. Accessibility is not only about compliance but also about clarity. If users don’t understand what is happening, they lose trust in the software. This is where clear error messages, loading states, upload feedback, etc. come in.</p><p>Pawel observed serendipitous side effects from improving accessibility, such as cleaner code, fewer bugs, and better UX. Thus he now thinks that accessibility is not a checklist but a mindset. If a software is accessible, it is likely to be engineered well.</p><p><strong>“A Walkthrough of RAG” by Josh</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VREa1sN90BSlM14BnyAGNg.jpeg" /><figcaption>Josh presenting “A Walkthrough of RAG”</figcaption></figure><p>Retrieval-Augmented Generation (RAG) is a process which involves optimizing the output of a Large Language Model (LLM). While most LLMs are trained on very large datasets, they do give inaccurate information every now and then. RAG tries to solve this by upserting additional data to help prevent LLMs from hallucinating.</p><p>In simpler terms, RAG works by allowing the LLM to trigger a search in the Vector Database. Retrieved relevant information is put back into the prompt as context. Vector Databases store, index, and search high-dimensional numeric representations of data. This enables rapid similarity search by finding close data points. Vector database also enables semantic search, which is an advanced search technique that understands the meaning of users’ queries. Semantic search uses natural language processing (NLP) and machine learning (ML) to understand context and relationship between words.</p><p>Some common use cases for RAG include customer service chatbots, healthcare diagnosis assistance, legal research, financial analysis, and code generation.</p><p><strong>“Catching Up With Stranger Things Using RAG” by Ryo</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*F_Qx5unK0mPR4TtvM4zARw.jpeg" /><figcaption>Ryo presenting “Catching Up With Stranger Things Using RAG”</figcaption></figure><p>Ryo tried to use RAG to catch up on the first 4 seasons of Stranger Things in preparation of its final season. There are multiple years between seasons, so he looked to RAG to help him keep track of the story.</p><p>Ryo’s RAG implementation utilized <a href="https://aws.amazon.com/bedrock/">Amazon Bedrock</a>, not the least of which its provided web crawler to retrieve relevant information. He also utilized <a href="https://aws.amazon.com/opensearch-service/features/serverless/">Amazon OpenSearch Serverless</a> as the vector database, i.e., storing the information retrieved by the web crawler.</p><p>Unfortunately, Ryo thinks this use case is not exactly the best usage of RAG, considering existing tools such as ChatGPT and Gemini already has access to such information. A better example would be LLMs that work with data that is not as widely available.</p><p><strong>“How The Beatles Innovated Music Technology” by Luke</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kfXpuhuWxKO71lmDmvVd0w.jpeg" /><figcaption>Luke presenting “How The Beatles Innovated Music Technology”</figcaption></figure><p>The music of The Beatles progressed very quickly. Their first album, consisting of 14 tracks, took 26 hours to record. On the other hand, their last, consisting of 17 tracks, took a whopping 379.</p><p>The technology The Beatles used in making their music progressed in a similarly rapid manner. Luke explained some of these technologies, the first of which is Artificial Double Tracking. It is a technique developed as a technical alternative to the tedium of double tracking. Tape delay is used to create a delayed copy of an audio signal, which is then played back at slightly varying speed (controlled by an oscillator) and combined with the original.</p><p>As a result of their continuous effort in experimenting with their music, The Beatles also participated in the invention of or popularized other techniques. Backmasking is a recording technique in which a message (e.g., music) is deliberately recorded backward onto a track that is meant to be played forward. To boost bass guitar sound, The Beatles used a loudspeaker as a microphone. To prevent the needles of old record players from jumping from really high bass, a tool called Automatic Transient Overload Control was developed in Abbey Studios. Finally, close miking is a technique The Beatles used to achieve really biting sounds of strings.</p><p><strong>“Introducing Town Hall: Aligning on Where We Are and What’s in Motion” by Vi and Mussin</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0WQ0I-gDE5e-xzeJfY_hzw.jpeg" /><figcaption>Vi and Mussin presenting “Introducing Town Hall: Aligning on Where We Are and What’s in Motion”</figcaption></figure><p>Vi and Mussin proposed town hall meetings as a way to more strategically allocate time for improvements across multiple teams. This is especially important as the number of people, teams, roles, and projects increase. For instance, if there are multiple projects that are trying to solve the same problem but from different angles, they should be consolidated. For another, everyone should agree on what are the most pressing issues that need to be addressed.</p><p>Town hall meetings promises to help everyone to be on the same page in this regard. It is to be used to clarify what is currently in motion, to align as to what the most pressing issues are, to come up with ways to resolve those issues strategically, and to learn from each other. To that end, the topics to be discussed in town hall meetings are those that concern multiple teams and/or projects, those that pertain to resources, and those that are time-sensitive.</p><p><strong>“Process in Action — Part 9” by Doi</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wtb0tX0nvRkIXjaxogaE_g.jpeg" /><figcaption>Doi presenting “Process in Action — Part 9”</figcaption></figure><p>Doi-san returned for the ninth installment of his Process in Action series. This time around, he focused on explaining the merits of managing contracts and tenants separately. If nothing else, this approach should lead to more agility and more scalability. It may seem unintuitive considering there will be 2 management processes to handle instead of 1, but the agility and scalability should come from the separation of concern between them.</p><p>Regarding contracts, Doi-san also explained the idea of agreement and entitlement being different concepts. Agreement is the commercial truth, focusing on billing, pricing, and contract duration. Entitlement is the technical truth, focusing on provisioning, API limits, and feature tiers. Among others, this idea has been adopted by AWS Marketplace, what with its separate definitions of <a href="https://docs.aws.amazon.com/marketplace/latest/userguide/viewing-your-agreements.html">agreement</a> and <a href="https://docs.aws.amazon.com/marketplace/latest/userguide/checking-entitlements.html">entitlement</a>.</p><p>As always, we had a party afterwards 😄</p><p>See you at the next MTS!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a0b39ce4a698" width="1" height="1" alt=""><hr><p><a href="https://medium.com/henngeblog/137th-monthly-technical-session-a0b39ce4a698">137th Monthly Technical Session</a> was originally published in <a href="https://medium.com/henngeblog">henngeblog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[134th Monthly Technical Session]]></title>
            <link>https://medium.com/henngeblog/134th-monthly-technical-session-2520a1bc4319?source=rss-c591be081b7d------2</link>
            <guid isPermaLink="false">https://medium.com/p/2520a1bc4319</guid>
            <category><![CDATA[monthly-technical-session]]></category>
            <category><![CDATA[figma]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[llm]]></category>
            <category><![CDATA[devops]]></category>
            <dc:creator><![CDATA[Bagus Aryabima]]></dc:creator>
            <pubDate>Thu, 05 Mar 2026 02:01:01 GMT</pubDate>
            <atom:updated>2026-03-05T02:01:01.066Z</atom:updated>
            <content:encoded><![CDATA[<p>HENNGE’s Monthly Technical Session (MTS) is a platform for members to share knowledge with their peers. Each speaker gives a talk for 10–20 minutes, followed by a Q&amp;A session. Occasionally there is also a lightning talk session at the end of an MTS, in which speakers give 5-minute talks.</p><p>The 134th MTS was held on September 26, 2025.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*2dPNdqY99zbIgNucb-Qkuw.jpeg" /><figcaption>Nektar being the MC of the 134th MTS</figcaption></figure><p><strong>“Playing Around With the Katalon AI Assistant” by Ota-san</strong></p><p><a href="https://katalon.com/">Katalon Studio</a> is a comprehensive automation testing solution that supports testing web, mobile, and APIs. It enables users to create, execute, and manage test cases with ease. Katalon Studio Assist is a built-in intelligent assistant within Katalon Studio that provides users with contextual help, coding suggestions, and automated test script generation to streamline the testing process. Katalon’s AI explicitly does not learn from the content it reads.</p><p>Ota-san has been using Katalon Studio as well as Katalon Studio Assist for a while, and she demonstrated some of the latter’s features: code generation and code explanation. Katalon Studio Assist is able to generate code for scenarios such as logging in to a web application and check whether certain cookies are issued afterwards. Katalon Studio Assist is also able to explain code (including those it generated) that should, at the very least, provide hints for beginners.</p><p><strong>“Under the Hood: Dissecting a Container Image” by Corrina</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nzdmZLg_rxLFzARobZcYzw.jpeg" /><figcaption>Corrina presenting “Under the Hood: Dissecting a Container Image”</figcaption></figure><p>A container is fundamentally a lightweight VM, which reuses parts of the operating system to lower overhead. Corrina went into depth regarding container images, dissecting one and explaining the various parts of the image. Unzipping a custom-built Docker image using the tar command revealed 3 primary files and 1 directory. index.json is the “main” entry point. The oci-layout file indicates that the image is an <a href="https://opencontainers.org/">OCI</a> image. OCI is the most commonly used standard for container images. manifest.json is provided for backwards compatibility with Docker, i.e., it is not OCI-compliant. index.json is the “main” entry point.</p><p>And then there is the blobs directory. Files within this directory are content-addressable, i.e., the SHA-256 checksum of a file equals its filename. Otherwise, it means the file has been modified. Corrina followed the manifests found in index.json and files within the blobs directory, which revealed some details into how the image was built to begin with. Furthermore, by unzipping all the layers of the image on top of each other, it is possible to form the final version of the filesystem. Ultimately, a container image is simply an image of a Linux filesystem (or a Windows one).</p><p>Once the filesystem is unzipped, Corrina ran some security tools on it to see if vulnerabilities could be found. Using a tool like <a href="https://trivy.dev/">Trivy</a>, it is possible to create a Software Bill of Materials (SBOM) from the filesystem. SBOM catalogs the contents of the filesystem, including details about the operating system, its version, and installed packages. Trivy was then used again, this time to compare the SBOM against its vulnerability database.</p><p><strong>“Becoming the Right Amount of Mad at Your Online Experience” by Jon</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*qBhAaWkTr15I3xHgEuisnw.jpeg" /><figcaption>Jon presenting “Becoming the Right Amount of Mad at Your Online Experience”</figcaption></figure><p>A low-information diet is a strategy to intentionally limit distracting, unnecessary, or negative information to free up mental energy for more productive or meaningful activities, boosting focus and well-being. In this talk, Jon shared with us some of the practical things he has tried. A big idea shared among his approaches is reintroducing friction that platforms have removed. Another is finding ways to eliminate addictive hooks from social media and other platforms. Sometimes, pursuing these 2 ideas involve using tools in unintended ways. For example, when dealing with X, Jon eventually found that resetting but not saving his password created a sufficient barrier to keep him off the site.</p><p>Obviously, different platforms remove different frictions and have different addictive hooks, so oftentimes each platform requires a specific approach. For Reddit, Jon recommended using <a href="https://www.firefox.com/en-US/">Firefox</a> browser and downloading <a href="https://ublockorigin.com/">uBlock Origin</a> extension. While the browser extension is primarily for blocking ads, it can also be used to block arbitrary elements on a webpage.</p><p>The title of the talk refers to a practice that unexpectedly helped Jon’s efforts, which is to categorize emotions by the strength of the trigger and the anger in response. To visualize this, Jon created a Punnett square for both aspects. Since users ultimately cannot control what the platforms they are using do, it is important to use tools such as Jon’s Punnett square to tweak their responses. The right amount of mad refers to responses which are proportional to the triggers.</p><p><strong>“Config 2025: A Very Short Report From Figma’s Annual Conference” by Mishela and Tina</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AgShYc1MiTGhd6-BgnnXXw.jpeg" /><figcaption>Mishela and Tina presenting “Config 2025: A Very Short Report From Figma’s Annual Conference”</figcaption></figure><p>Mishela and Tina presented a report on <a href="https://www.figma.com/blog/config-2025-recap/">Config 2025</a>, which is Figma’s annual conference for product builders, including designers, developers, and marketers. The conference was held in San Francisco as well as online and spanned 2 days of talks, presentations, and networking opportunities. It also features keynotes and sessions where Figma shares its latest updates and releases, some notable ones being <a href="https://www.figma.com/buzz/">Figma Buzz</a>, <a href="https://www.figma.com/draw/">Figma Draw</a>, <a href="https://www.figma.com/sites/">Figma Sites</a>, <a href="https://www.figma.com/make/">Figma Make</a>, and <a href="https://help.figma.com/hc/en-us/articles/31289469907863-Use-the-grid-auto-layout-flow">Figma Grid.</a></p><p>Some of the talks that Tina found most interesting are <a href="https://www.youtube.com/watch?v=Akudd_qFWrg">Designing in Disaster</a>, <a href="https://www.youtube.com/watch?v=nIq4pHAIEeU">Designing for Play and Friction in a Fast-Paced World</a>, and <a href="https://www.youtube.com/watch?v=dJ4DoFzTcBc">Redesigning the Computer Mouse</a>. For Mishela, the talks she found most interesting are <a href="https://www.youtube.com/watch?v=zxaHuTyaVfQ">Trust is the Currency</a>, <a href="https://www.youtube.com/watch?v=nIq4pHAIEeU">How To Be a Robot Whisperer</a>, and <a href="https://www.youtube.com/watch?v=bbZTRsvmzbQ">The Space-Filling Curve of Design</a>.</p><p>A major theme of the conference, by far was AI. Talks about AI generally leaned towards the belief that AI would be the answer to everything or focused on casual use, with few discussions about the repercussions of AI. For its part, the conference also stressed the importance of honing our craft and developing strong design judgement. At the end of the day, Figma is a tool, and what truly matters is building skills that transcend any specific software.</p><p><strong>“Create Your Own LLM Chat Interface” by Yoel</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1ttrbYKKgtz_l4I1_vRgvQ.jpeg" /><figcaption>Yoel presenting “Create Your Own LLM Chat Interface”</figcaption></figure><p>In the era of large language models (LLMs), users often face hurdles such as being restricted to a limited selection of models or incurring high costs from monthly subscriptions. Yoel introduces a solution, which is building your own self-hosted LLM chat interface. By leveraging open-source tools and a specialized proxy architecture, this solution grants the freedom to experiment with a wider array of models while gaining better control over their usage and expenditure.</p><p>Yoel’s LLM chat interface relies on a few key components. <a href="https://github.com/open-webui/open-webui">OpenWebUI</a> serves as the self-hosted AI platform and user interface, offering features like multi-model support and customizability through pipelines and plugins. To connect OpenWebUI to various LLM providers, 2 adapter layers are used: <a href="https://github.com/BerriAI/litellm">LiteLLM</a> (i.e., a self-hosted LLM API adapter) and an API gateway. Both adapters are crucial for exposing an OpenAI-compatible API, and they offer essential features for enterprise use, such as cost management, rate limiting, logging, and custom routing to optimize performance and control expenses.</p><p>Ultimately, deploying a local LLM platform provides significant advantages over relying solely on commercial subscriptions. It offers access to diverse LLM providers, ensures a more privacy-friendly environment through zero data retention on API-based calls, and provides a potentially cheaper alternative to fixed monthly subscriptions by shifting costs to a usage-based model. A solution like Yoel’s empowers users to tailor their LLM experience to their specific needs, both in terms of functionality as well as cost.</p><p><strong>“Software Engineer Tries Hardware — the Final Result!” by Jeanie</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-77QvU2Rf2O1UXOsEICE3Q.jpeg" /><figcaption>Jeanie presenting “Software Engineering Tries Hardware — the Final Result!”</figcaption></figure><p>Jeanie has been working on a hardware project for a while, and she has finally finished it. The project is about creating electronic badges, and she <br>created several iterations. Version 2 saw more capacitive touch pads, battery reverse polarity protection, and the use of electroless nickel immersion gold (ENIG). Version 3 introduced additional circuits to fix potential battery drain issue. The production version adjusted the LED location for aesthetic purposes as well as changed charging resistor.</p><p>As for the cases of the electronic badges, Jeanie went with injection molding because 3D printing — a more affordable approach — unfortunately did not work for her use case. After some more testing, the production version of the electronic badges were able to be delivered to 75 people.</p><p><strong>“DevSecOps Updates” from Habibi and Hans</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*LRiC2H5weJRVKMX1LjQ0uQ.jpeg" /><figcaption>Habibi and Hans presenting “DevSecOps Updates”</figcaption></figure><p>Habibi and Hans shared how they address security vulnerabilities and coding errors found by GitHub <a href="https://docs.github.com/en/code-security/code-scanning/introduction-to-code-scanning/about-code-scanning">code scanning</a>. Some of these were resolved by dismissing the alert, e.g., when they occur only in tests. But these findings vary quite a lot and each type requires a different approach to resolve. For example, some findings related to use of external libraries were resolved by upgrading the libraries, while some findings related to GHA workflow permissions were resolved by setting permissions explicitly.</p><p>Some findings were complex enough that resolving them requires the involvement of a whole team. In these cases, Habibi and Hans created dedicated sessions to go through the whole list, resolving them one by one from the least severe to the most.</p><p>As always, we had a party afterwards 😄</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jks4k5b9wo8nq9P8z6wMOg.jpeg" /><figcaption>MTS #137 Afterparty</figcaption></figure><p>See you at the next MTS!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2520a1bc4319" width="1" height="1" alt=""><hr><p><a href="https://medium.com/henngeblog/134th-monthly-technical-session-2520a1bc4319">134th Monthly Technical Session</a> was originally published in <a href="https://medium.com/henngeblog">henngeblog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[131st Monthly Technical Session]]></title>
            <link>https://medium.com/henngeblog/131st-monthly-technical-session-cbcb27f929e7?source=rss-c591be081b7d------2</link>
            <guid isPermaLink="false">https://medium.com/p/cbcb27f929e7</guid>
            <category><![CDATA[monthly-technical-session]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[dynamodb]]></category>
            <category><![CDATA[henngeculture]]></category>
            <dc:creator><![CDATA[Bagus Aryabima]]></dc:creator>
            <pubDate>Fri, 14 Nov 2025 02:02:25 GMT</pubDate>
            <atom:updated>2025-11-14T02:02:25.427Z</atom:updated>
            <content:encoded><![CDATA[<p>HENNGE’s Monthly Technical Session (MTS) is a platform for members to share knowledge with their peers. Each speaker gives a talk for 10–20 minutes, followed by a Q&amp;A session. Occasionally there is also a lightning talk session at the end of an MTS, in which speakers give 5-minute talks.</p><p>The 131st MTS was held on June 27th, 2025.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0CIqEHJ2JsTrmvdvP9BB1A.jpeg" /><figcaption>Rob being the MC of the 131st MTS</figcaption></figure><h4>“Beware the Gadgetbahn” by Jonas</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tbTT9DK-ZlaoPMkywob3JA.jpeg" /><figcaption>Jonas presenting “Beware the Gadgetbahn”</figcaption></figure><p>Jonas likes trains. He doesn’t like gadgetbahns as much. Gadgetbahn is a portmanteau of the English word “gadget” and the German word “bahn” (i.e., railway). It refers to public transportation that are touted as innovative but in practice is not necessarily better than traditional ones (e.g., trains).</p><p>For more clarity, Jonas compares trains with a type of gadgetbahn: monorails. Trains use steel wheels and run on steel rails. They are also motorized, electrified, and uses gauges that are standard to the area they run on. Trains are a tried and true public transportation.</p><p>Monorails promise to be a cheaper and innovative public transportation, but in reality they come with their own set of challenges. First, monorails have to be elevated. The concrete structure they require are expensive and does not allow for level crossings. If something goes wrong, the elevation makes evacuation and reparation difficult and dangerous. Second, monorail switches are complex, requiring a lot of energy and time to move. Third, monorails lack capacity. <a href="https://en.wikipedia.org/wiki/Line_3_(Chongqing_Rail_Transit)">CRT Line 3</a> is the world’s busiest monorail line with a daily ridership of about 600,000. <a href="https://en.wikipedia.org/wiki/T%C5%8Dyoko_Line">Toyoko Line</a>, which is just one of the many train lines in Tokyo, has a daily ridership of about 1,200,000. Fourth, monorail often leads to vendor lock-in. In fact, JR East is in the process of replacing <a href="https://en.wikipedia.org/wiki/Tokyo_Monorail">Tokyo Monorail</a> with a normal train line to enable connections to other existing rail networks.</p><p>After learning about these challenges, Jonas looked into the potential reasons why gadgetbahns are built to begin with. One may be lack of understanding. Decision makers may not understand public transportation well enough or may be swayed by persuasive salespeople without fully understanding the costs and benefits. Another is a desire for something “futuristic”, i.e., that is not an “old-fashioned” train. There have been cases in which this resulted in the adoption of flashy but impractical technologies. Yet another is conflict of interest. Entities with vested interest may propose suboptimal public transportation to distract from and undermine the development of effective public transportation.</p><p>Jonas emphasized that public transportation matter because it provides mobility for all people, especially those unable to drive. Public transportation also have long-term consequences. They persist for decades and impact future generations. Gadgetbahn oftentimes have more negative consequences than positive ones. It is also important to note that the decision makers are not always the ones affected by poor decisions.</p><p>But Jonas is not opposed to innovation. Rather, his point is that innovation has to be based on sufficient understanding, of both the innovation as well as its potential impact. One case of an innovation done right is <a href="https://en.wikipedia.org/wiki/Ch%C5%AB%C5%8D_Shinkansen">Chuo Shinkansen</a>, a maglev train being developed by JR Central between Tokyo and Nagoya. It is not technically a train, as it hovers rather than using wheels and rails most of the time. However, it is not a gadgetbahn either, due to being justified by an actual need: the existing line, i.e., Tokaido Shinkansen, is at capacity. Chuo Shinkansen being significantly faster than Tokaido Shinkansen also gives people a good reason to use the new line. Furthermore, this innovation is being developed by an entity with a long track record in rail technology.</p><p>Jonas concluded by advising innovators to start by picking the standard. Consider the context in which you build the system, and adapt to it. If there is an actual need to deviate, then proceed with sufficient understanding, of both the deviation and its impact.</p><h4>“What Happens When You Go Schemaless?” by Dima</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*b5Wj9DTnaKRQohpF2xvlcA.jpeg" /><figcaption>Dima presenting “What Happens When You Go Schemaless?”</figcaption></figure><p>Dima delves into the architecture of <a href="https://www.amazon.science/publications/dynamo-amazons-highly-available-key-value-store">Dynamo</a>, Amazon’s highly available key-value store. The talk is structured around 4 key aspects of its infrastructure: partitioning, versioning, consistency, and fault tolerance.</p><p>In key-value stores, each key describes the partition in which data is stored. Hash functions are used to map keys to partitions. Key-value stores are highly partitionable and allow horizontal scaling at a level that other types of databases cannot achieve. This means that partitions are frequently added or removed as needed, which proves to be a challenge for traditional hash functions such as MD5 or SHA-1. To address this, Dynamo utilizes <strong>consistent hashing</strong>. In this approach, a hash space is visualized as a ring, and each node is assigned a token representing a position on this ring. When a key is hashed, its position on the ring is determined, and the data is stored on the next node clockwise. Adding or removing nodes have minimal impact, as only a small portion of the data needs to be remapped. The lookup for the next node is performed efficiently using binary search, as all tokens on the hash ring are sorted.</p><p>Dynamo prioritizes high availability and allows for concurrent writes in a distributed environment, making managing data versions and reconciling conflicting ones crucial. Traditional timestamps are not suitable due to potential clock drifts and concurrency issues. Instead, Dynamo employs <strong>vector clocks</strong>. Each data object stores not only its value but also a vector clock, which records the operations performed by each server on that object (i.e., reads or writes). The client receives all possible versions of the data, even if they were written concurrently and the clients were not aware of each other. Conflicting versions are then reconciled by examining the vector clocks and merging them. This involves looking for the highest counts for each server in the vector clocks to create a reconciled version.</p><p>Dynamo forgoes strict consistency in favor of <strong>eventual consistency</strong>. While data may not be immediately consistent across all replica partitions after a write, it will eventually converge. To this end, Dynamo implements the formula <strong><em>R + W &gt; N</em></strong>, where <em>R</em> is the maximum number of replica partitions that must respond to a read operation, <em>W</em> is the minimum number of replica partitions that must respond to a write operation, and <em>N</em> is the total number of replica partitions. This formula ensures that there is at least one shared replica between read and write operations, thereby allowing for correct transactions to be performed despite eventual consistency.</p><p>Dynamo handles both temporary and permanent failures. Temporary failures are managed using <strong>hinted handoff</strong>. If partition A is temporarily unavailable, its data is sent to partition B with a hint that the data is intended for partition A. Once partition A is available again, partition B hands off the data back, resolving the temporary failure. Permanent failures are managed by detecting inconsistencies between replica partitions and then restoring the healthy state of the system. Each replica partition stores its own Merkle Tree, where data versions are hashed and then combined with parent hashes up to a root hash. Replica partitions send root hashes to each other (as opposed to full data), reducing latency. If hashes differ, the system navigates down the Merkle Trees to pinpoint the exact location of inconsistencies.</p><h4>“Chaos Engineering” by Eri</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aP42jvENPfcIZ9T8im2qVA.jpeg" /><figcaption>Eri presenting “Chaos Engineering”</figcaption></figure><p>Chaos engineering refers to intentionally injecting failures into a system in a controlled manner, in order to observe its behavior and pinpoint potential weaknesses and vulnerabilities. It involves people, tools, and practices, one of which is chaos experiments. Chaos experiments involve determining the steady state, defining scenarios of what could go wrong, running experiments (and monitoring performance metrics), fixing (or rolling back) and analyzing results, and finally repeating with a larger blast radius.</p><p>There have been big adopters of chaos engineering such as Netflix and Wix, but Eri wonders if it is still a thing. On one hand, most resources about chaos engineering are dated. On the other hand, today there is a long list of tools that can be used to easily run chaos experiments and build resilience, such as <a href="https://chaos-mesh.org/">Chaos Mesh</a>, <a href="https://www.gremlin.com/">Gremlin</a>, <a href="https://spinnaker.io/">Spinnaker</a>, and others. Eri was able to find about 500 chaos engineering tool repositories on GitHub. Something else to consider is that learning from past mistakes has led to a greater degree of resilience being baked into resources than before.</p><p>For her part, Eri considers chaos engineering to still be a thing. It has matured, and adopting it is easier than before. There are several ways to start doing so, such as doing gamedays (i.e., to simulate failure events) and building a culture (e.g., starting on an opt-in basis early). However, it is important to remember that depending on the chaos, running scheduled experiments can be costly, dangerous, and perhaps not even realistic to begin with.</p><h4>“AI Agent Application Development” by Jack</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9OUTultYK9nikWuadwRy5Q.jpeg" /><figcaption>Jack presenting “AI Agent Application Development”</figcaption></figure><p>AI agents are essentially large language model (LLM) wrappers designed to reduce hallucinations. They operate on the “Reason and Act” (ReAct) paradigm, where an LLM thinks about its own response, plans it, executes it, and then reevaluates the results in an iterative loop until it reaches a desired answer. A key aspect of AI agents is their ability to call tools, allowing them to interact with underlying architectures such as databases.</p><p>AI agents can independently solve practical and complex problems as well as generalize tasks. The latter is a key difference between AI agents and automated workflows. AI agents can handle similar but varied tasks without manual setup of an automation pipeline. This capability is seen as a breakthrough for businesses, particularly for Software as a Service (SaaS) companies.</p><p>AI agent tech stack spans several areas, such as model serving &amp; inference (e.g., <a href="https://gemini.google/overview/">Gemini</a>, <a href="https://ollama.com/">Ollama</a>), storage (e.g., <a href="https://www.pinecone.io/">Pinecone</a>, <a href="https://www.trychroma.com/">Chroma</a>), agent development (e.g., <a href="https://aws.amazon.com/bedrock/">Bedrock</a>, <a href="https://www.langchain.com/langgraph">LangGraph</a>), observability (e.g., <a href="https://www.langchain.com/langsmith">LangSmith</a>, <a href="https://www.comet.com/site/products/opik/">Opik</a>), tool execution (e.g., <a href="https://exa.ai/">Exa</a>, <a href="https://composio.dev/">Composio</a>), and memory management (e.g., <a href="https://www.langchain.com/langchain">LangChain</a>). In particular, the agent development frameworks wrap the ReAct paradigm loop and tool-calling functions. Each framework has its own logic for querying the tools to be called and performing their own reasoning.</p><p>Jack presented a demo of a project where an AI agent interacts with network data to provide insights. The AI agent calls graph-building tools to render charts and calls SQL tools to query the database. The idea is to not require users to navigate all around a monitoring solution to gain insights. With this approach, users simply type in the insights they are looking for.</p><p>Jack ended his talk by highlighting the current issues with AI agents. There is a trade-off between cost and effort size in improving reliability. Two ways to achieve this are to use more expensive LLMs or implementing more sophisticated logical frameworks on top of existing LLMs. Whether LLMs should be trusted to handle complicated data at all is another concern. Yet another one is safety. For example, some AI agents query databases. It is important to limit the statements that AI agents can use, and even then there may still be queries that should not be executed. Finally, the time it takes for an AI agent to run and provide an accurate response impacts user experience.</p><h4>“Haptic Feedback in VR” by Luca</h4><p>Haptic technology refers to technology that can create an experience of touch by applying forces, vibrations, or motions to the user. It can be used to provide sensory information to improve accessibility and immersion.</p><p>There exist many types of haptic feedback, such as force feedback (e.g., resistance in racing wheels), electrical muscle simulation, vibrotactile (i.e., using vibrations to convey information), and many more. Vibrotactile is one of the most ubiquitous type due to its simplicity, affordability, and portability. It is common in commercial electronics such as smartphones and game controllers. There are at least 2 types of vibrotactile actuators: eccentric rotating mass (ERM) and linear resonant actuator (LRA). The former is cheap and simple, while the latter is more responsive and expressive.</p><p>The goal of virtual reality (VR) is to simulate reality, which is multisensory, in a virtual environment. Luca explained that there are 8 known human senses: visual (seeing), auditory (hearing), olfactory (smelling), gustatory (tasting), tactile (touching), vestibular (balance and spatial orientation), proprioception (position and movement), and interoception (internal state of the body). In contemporary VR, there have already been significant advancements in the visual and auditory domains. By leveraging haptic feedback, it is possible to target different senses (i.e., tactile, vestibular, and proprioception) to create a more immersive multisensory experience.</p><p>Luca has worked on leveraging haptic illusions to trick the brain into feeling more than what is physically present, i.e., resulting in a deeper immersion in VR. His approach was to use vibrations on the head to simulate the sensation of movement. This applies vibrotactile feedback to the head as well as tapping into the vestibular system. Furthermore, there are no treadmills or motion platforms needed. Instead, Luca created a wearable headband with LRAs in a 360 degrees layout.</p><h4>“Introduction of Partner Sales Section” by Michika and Hiromi</h4><p>To win together is one of the <a href="https://recruit.hennge.com/en/culture/">HENNGE WAY</a>s. To that end, Michika and Hiromi introduced the Partner Sales Section that they belong to. By knowing what Partner Sales Section does, other sections can ask their help regarding relevant things and vice versa. Michika and Hiromi explained how HENNGE One is proposed to potential customers and the strategies involved in working together with resellers.</p><h4>“Software Engineer Trying Hardware Part 2 — First Boards Are Here!” by Jeanie</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xpTDGakEIFcXlSFi9JBO9Q.jpeg" /><figcaption>Jeanie presenting “Software Engineer Trying Hardware Part 2 — First Boards Are Here!”</figcaption></figure><p>Jeanie has been working on a hardware project. The first step she had to do is coming up with the schematics for the hardware. To identify the required parts to be put in the schematics, she referred to sample circuits and checked the suggestions for resistors or capacitors to use.</p><p>The second step is designing the layout. The parts in the schematics are imported into a space in which she had to navigate and connect one part to another across multiple layers.</p><p>The third step is ordering the boards. Jeanie also gave us some insights of the economics of this process. Some sites may require a minimum amount of board to place an order. Generally, the more boards that are ordered, the lower the price per board is.</p><p>The fourth step is testing the boards. This is when she realizes what she forgot, what she might want to change, and what she might want to add.</p><p>The fifth step is revising the schematic and layout based on the test results. This is followed by reordering the boards.</p><p>The last step Jeanie was working on was learning 3d modelling. This would allow her to design an enclosure for the boards.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*raYi4m5rofmeTXXRe5TY7w.jpeg" /><figcaption>Jeanie and friends showcasing the boards she is working on</figcaption></figure><p>As always, we had a party afterward 😄</p><p>See you at the next MTS!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cbcb27f929e7" width="1" height="1" alt=""><hr><p><a href="https://medium.com/henngeblog/131st-monthly-technical-session-cbcb27f929e7">131st Monthly Technical Session</a> was originally published in <a href="https://medium.com/henngeblog">henngeblog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[128th Monthly Technical Session]]></title>
            <link>https://medium.com/henngeblog/128th-monthly-technical-session-446fa3aa5b06?source=rss-c591be081b7d------2</link>
            <guid isPermaLink="false">https://medium.com/p/446fa3aa5b06</guid>
            <category><![CDATA[tech]]></category>
            <category><![CDATA[vibe-coding]]></category>
            <category><![CDATA[japan]]></category>
            <category><![CDATA[monthly-technical-session]]></category>
            <category><![CDATA[hennge]]></category>
            <dc:creator><![CDATA[Bagus Aryabima]]></dc:creator>
            <pubDate>Fri, 11 Jul 2025 02:01:43 GMT</pubDate>
            <atom:updated>2025-08-19T03:07:45.741Z</atom:updated>
            <content:encoded><![CDATA[<p>HENNGE’s Monthly Technical Session (MTS) is a platform for members to share knowledge with their peers. Each speaker gives a talk for 10–20 minutes, followed by a Q&amp;A session. Occasionally there is also a lightning talk session at the end of an MTS, in which speakers give 5-minute talks.</p><p>The 128th MTS was held on March 28, 2025.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AHQJWRukOCiZB7lqEmxtXw.jpeg" /><figcaption>Rob being the MC of the 128th MTS</figcaption></figure><p><strong>“Getting Started With Renovate” by Habibi</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Prhb7F3rq0nVmw2UwydJrA.jpeg" /><figcaption>Habibi presenting “Getting Started With Renovate”</figcaption></figure><p>Habibi opened the 128th MTS by introducing <a href="https://docs.renovatebot.com/">Renovate</a>, an automated dependency update tool. In HENNGE, there are so many dependencies that it’s impossible for a human to update them one by one, and vulnerable ones have to be updated as soon as possible.</p><p>Renovate improves this situation by automatically creating dependency update PRs. The creation of these PRs can be scheduled very flexibly, and they happen instantaneously for vulnerable dependencies. Dependency dashboard provides a birds eye view of the various states of dependency updates, i.e., awaiting schedule, pending approval, rate limited, edited/blocked, open, ignored/blocked, etc. To top it all off, Renovate supports all package managers one can think of.</p><p>Habibi also explained how Renovate is run at HENNGE and invited us all to use it. To sweeten the deal, he also went through some of the more advanced features of Renovate, such as GitHub Actions hash pinning, Dockerfile hash pinning, custom manager with regex, grouping, and upgrading EKS version. Finally, he showcased the newest feature his team came up with: Renovate Job Logs API. Accessing the logs promises to ease the debug process of dependency updates done by Renovate.</p><p><strong>“A Brief Overview of AWS (in Anime Terms)” by Luke</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*S-tWd5eftbfpB_FmXOAs8A.jpeg" /><figcaption>Luke presenting “A Brief Overview of AWS (in Anime Terms)”</figcaption></figure><p>Luke aims to give a basic, high-level overview of <a href="https://aws.amazon.com/">Amazon Web Services</a> (AWS) in an interesting way by making analogies between them and anime. He likened AWS to a universal energy source that gives users the ability to build and scale their applications. Mastery over this energy source determines how strong one is in the digital world.</p><p>Like most anime, his explanations are divided into arcs:</p><ul><li>Training arc: first deployment</li><li>Serverless arc: exploring the cool parts of AWS and finding out why they are important</li><li>Tournament arc: tackling several common battles in AWS</li></ul><p>The training arc covers the basic techniques with which one can make pretty much anything: <a href="https://aws.amazon.com/s3/">Amazon S3</a>, <a href="https://aws.amazon.com/ec2/">Amazon EC2</a>, <a href="https://aws.amazon.com/elasticloadbalancing/">Elastic Load Balancing</a> (ELB), and <a href="https://aws.amazon.com/rds/">Amazon Relational Database Service</a> (RDS). S3 is an object storage built to retrieve any amount of data from anywhere. It is not unlike the Unlimited Void technique from Jujutsu Kaisen, which is a boundless domain where everything exists and can be accessed instantly. EC2 is a scalable cloud computing service that provides virtual servers (i.e., instances) to run applications. Its customizability reminds Luke of Gundams. ELB distributes network traffic to improve application stability, much like commanders in Attack on Titan leading their soldiers into battle (?). RDS is a managed service that makes it easier to set up, operate, and scale relational databases in the cloud. Luke admitted that this is a stretch, but the way RDS sets its users up for success reminded him of how Haikyu!!’s Tobio Kageyama sets a volleyball to his teammates.</p><p>The serverless arc is all about running applications without managing servers. Luke highlighted several services related to the concept: <a href="https://aws.amazon.com/lambda/">AWS Lambda</a>, <a href="https://aws.amazon.com/api-gateway/">Amazon API Gateway</a>, and <a href="https://aws.amazon.com/dynamodb/">Amazon DynamoDB</a>. Lambda is a serverless computing service that lets you run code without provisioning or managing servers. Your Lambda functions runs only when triggered by events, similar to how Lelouch’s Geass activates only when he looks into someone’s eyes. API Gateway is a fully managed service that allows you to create, manage, and secure APIs. Luke made an analogy between the service and the Hyperbolic Time Chamber from Dragon Ball, what with both controlling who is allowed inside and limiting how many entities can enter at a time. DynamoDB is a serverless, NoSQL, fully managed database with single-digit millisecond performance at any scale. It is like a magical library in various animes that automatically expands to accommodate infinite books, allowing one to quickly retrieve and store books.</p><p>Finally, the tournament arc sheds some light on permissions, cost, and security. Luke gave us the keys to victory in all 3 battles:</p><ul><li>Permission: principle of least privilege</li><li>Cost: use the tools (e.g., <a href="https://aws.amazon.com/aws-cost-management/aws-cost-explorer/">AWS Cost Explorer</a>, <a href="https://aws.amazon.com/aws-cost-management/aws-budgets/">AWS Budgets</a>, and <a href="https://docs.aws.amazon.com/cur/latest/userguide/what-is-cur.html">AWS Cost and Usage Reports</a>)</li><li>Security: principle of least privilege (e.g., in <a href="https://docs.aws.amazon.com/vpc/latest/userguide/vpc-security-groups.html">security groups</a>) <em>and</em> use the tools (e.g., <a href="https://aws.amazon.com/waf/">AWS WAF</a>, <a href="https://aws.amazon.com/shield/">AWS Shield</a>, <a href="https://aws.amazon.com/kms/">AWS Key Management Service</a>, <a href="https://aws.amazon.com/certificate-manager/">AWS Certificate Manager</a>, <a href="https://aws.amazon.com/security-hub/">AWS Security Hub</a>, <a href="https://aws.amazon.com/guardduty/">Amazon GuardDuty</a>, <a href="https://aws.amazon.com/iam/features/mfa/">Multi-Factor Authentication for IAM</a>, etc.)</li></ul><p><strong>“#fun-skiing-snowboarding-bu 2025 Season Review” by Iggy</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JEGCwhyDDm_xaj9uM_ujug.jpeg" /><figcaption>Iggy presenting “#fun-skiing-snowboarding-bu 2025 Season Review”</figcaption></figure><p>HENNGE created its <em>Bukatsu (</em>Fun Activity Club) Program to encourage cross-divisional collaboration and cultivate off-the-work relationships to bind HENNGE members together. Iggy is part of the #fun-skiing-snowboarding-bu, which is a <em>Bu</em> (club) for winter sports lovers who plan trips together. During the 2025 season, #fun-skiing-snowboarding-bu planned 3 trips: to <a href="https://www.maiko-resort.com/">Maiko Resort</a>, to <a href="https://sugadaira-snowresort.com/">Sugadaira Kogen Snow Resort</a>, and to <a href="https://www.hakubavalley.com/en/">Hakuba Valley</a>.</p><p>Iggy was curious about whether #fun-skiing-snowboarding-bu fulfilled the objective of the <em>Bukatsu</em> Program. Out of the 5–12 people who went on trips during the 2025 season, it turns out most of them are from the same division. But there was always one person from other divisions, so cross-divisional collaboration was encouraged to a certain extent.</p><p>To improve division diversity, Iggy looked into the most popular <em>Bu</em>s and learned how they plan their events. Early notice seems to be important. The most popular <em>Bu</em>s announce their events at least 3-weeks in advance. Within #fun-skiing-snowboarding-bu itself, the trips that were announced earlier also had higher member turnout. The organizers’ language proficiency also plays a factor. Events organized by multilingual people are attended by 20 people or more.</p><p>Looking into the 2026 season, Iggy set a goal of making #fun-skiing-snowboarding-bu for all of HENNGE, not just a few divisions within the company. To achieve this, he plans to announce the trips earlier and collaborate with other divisions not only during the trips but also in their planning processes.</p><p><strong>“Linux Tool Alternatives” by Ogawa-san</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Dtls-30SAKI8N_PALf4V1g.jpeg" /><figcaption>Ogawa-san presenting “Linux Tool Alternatives”</figcaption></figure><p>There are many deprecated Linux tools, i.e., ones that are no longer actively maintained or updated. People still use deprecated tools for one reason or another, but these days, there are improved alternatives that provide the same functionality.</p><p>Ogawa-san showcased some of the modern Linux tools that he has been using. <a href="https://github.com/tldr-pages/tldr">tldr</a> is a collection of community-maintained help pages for command-line tools that aims to be a simpler, more approachable complement to traditional <a href="https://en.wikipedia.org/wiki/Man_page">man pages</a>. <a href="https://github.com/BurntSushi/ripgrep">ripgrep</a> is a line-oriented search tool that recursively searches the current directory for a regex pattern, and it is a very fast <a href="https://en.wikipedia.org/wiki/Grep">grep</a> replacement. <a href="https://github.com/sharkdp/fd">fd</a> is a program to find entries in your filesystem. It is a simple, fast, and user-friendly alternative to <a href="https://www.gnu.org/software/findutils/">find</a>. <a href="https://github.com/junegunn/fzf">fzf</a> is a general-purpose command-line fuzzy finder. <a href="https://github.com/httpie/cli">httpie</a> is a command-line HTTP client. Its goal is to make CLI interaction with web services as human-friendly as possible, making it a promising alternative to <a href="https://curl.se/">curl</a> and <a href="https://www.gnu.org/software/wget/manual/wget.html">wget</a>. Last but not least, <a href="https://github.com/sharkdp/bat">bat</a> is a <a href="https://en.wikipedia.org/wiki/Cat_(Unix)">cat</a> clone with syntax highlighting and Git integration.</p><p><strong>“Behind the Hood: Understanding Web Page Rendering Process in Three Levels of Complexity” by Lucas</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0iu34vrzeUYx6ruNLDTBqw.jpeg" /><figcaption>Lucas presenting “Behind the Hood: Understanding Web Page Rendering Process in Three Levels of Complexity”</figcaption></figure><p>Lucas divided the web page rendering process into 3 levels of complexity. On the basic level, webpages need to go into the Critical Rendering Path: the sequence of steps the browser goes through to convert the HTML, CSS, and JavaScript into pixels on the screen. There are at least 5 components involved on this level: network, byte stream decoder, input stream processor, tokenizer, and tree construction. According to Lucas, the last 2 are the most important. Tokenizers are state machines that take a stream of bytes and convert them into tokens. Tree construction is a process that is divided into 3 smaller components: the DOM, the CSSOM, and the render tree. The DOM is a tree-based data structure that is used to render HTMLs to the render tree. The CSSOM is similar to the DOM, albeit for CSSs. The DOM and the CSSOM are created asynchronously. Finally, both the DOM and the CSSOM are combined to create the render tree.</p><p>On the level of updating the content the user sees, this time, Lucas explained an approach that involves React. At the heart of React is Fiber, an internal data structure that manages state updates and virtual DOM creation. Another important part is the reconciliation algorithm, which has 2 phases: Render Phase (asynchronous, can be paused/stopped/cancelled, lower priority) and Commit Phase (synchronous, can not be paused/stopped/cancelled, higher priority). The Render Phase constructs new virtual DOMs in memory, essentially preparing the changes to be made. The Commit Phase sees the HostRoot shift its pointer from the real DOM to the virtual DOM (the very same one as created in the Render Phase), after which React updates the real DOM with changes marked in the virtual DOM.</p><p>On the level of requesting strategies, Lucas explained client-side rendering (CSR), server-side rendering (SSR), and incremental static regeneration (ISR). CSR has better interactivity, exposes actions to the client, and is good for highly dynamic content and real-time updates. SSR adds interactivity via hydration, fetches pages fully per request, and is good for SEO and content-heavy apps. Finally, ISR renders on both client and server, and despite the fact that its data can become stale, it is good for frequent data updates and blogs.</p><p><strong>“Vibe Coding” by Garv</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*PVsAdK7ODwuEDLTS_YGXHQ.jpeg" /><figcaption>Garv presenting “Vibe Coding”</figcaption></figure><p><a href="https://en.wikipedia.org/wiki/Vibe_coding">Vibe coding</a> is a term coined by Andrej Karpathy (Co-Founder of OpenAI and former AI leader at Tesla), which essentially means AI-dependent programming. It is an approach to producing software where a person describes a problem in a few sentences as a prompt to a large language model (LLM) tuned for coding, and the LLM generates software.</p><p>With vibe coding, building applications is now within reach of people who have never written a line of code. Using AI code editors like <a href="https://chatgpt.com/">Cursor</a> to reference your files and edit your code, and then using <a href="https://chatgpt.com/">ChatGPT</a> to bounce ideas off of are oftentimes all it takes to vibe code. This lets people focus on their visions, their ideas, and doing rapid iteration, which is something startups thrive on.</p><p>There have been several succesful stories of vibe coding. <a href="https://fly.pieter.com/">fly.pieter.com</a> is a game created with AI the span of 30 minutes to 3 hours, generating more than $50,000 a month at one point. <a href="https://www.calai.app/">Cal AI</a> tracks your calories with just a picture, and has now gotten over 100k 5-star ratings.</p><p>But vibe coding is not all sunshine and rainbows. <a href="https://resources.uplevelteam.com/gen-ai-for-coding">One research</a> suggests that developers with access to AI coding assistants saw a 41% increase in bug rate, while their issue throughput remained consistent. Vibe coding can also contribute to technical debt, i.e., the extra work required when developers choose an easy short-term solution over the best long-term approach. Furthermore, it has been shown that a considerable portion of AI-generated code contains security flaws and deprecated dependencies.</p><p>Circling back to the one who coined the term vibe coding, Andrej Karpathy concluded that the technique is “not too bad for throwaway weekend projects”. For his part, Garv concluded that it may not be wise to completely do vibe coding on production.</p><p>As always, we had a party afterward 😄</p><p>See you at the next MTS!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=446fa3aa5b06" width="1" height="1" alt=""><hr><p><a href="https://medium.com/henngeblog/128th-monthly-technical-session-446fa3aa5b06">128th Monthly Technical Session</a> was originally published in <a href="https://medium.com/henngeblog">henngeblog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[122nd Monthly Technical Session]]></title>
            <link>https://medium.com/henngeblog/122nd-monthly-technical-session-20b99b2e8913?source=rss-c591be081b7d------2</link>
            <guid isPermaLink="false">https://medium.com/p/20b99b2e8913</guid>
            <category><![CDATA[isms-certification]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[monthly-technical-session]]></category>
            <category><![CDATA[programming-languages]]></category>
            <dc:creator><![CDATA[Bagus Aryabima]]></dc:creator>
            <pubDate>Mon, 17 Feb 2025 02:02:25 GMT</pubDate>
            <atom:updated>2025-02-17T02:02:25.111Z</atom:updated>
            <content:encoded><![CDATA[<p>HENNGE’s Monthly Technical Session (MTS) is a platform for members to share knowledge with their peers. Each speaker gives a talk for 10–20 minutes, followed by a Q&amp;A session. Occasionally there is also a lightning talk session at the end of an MTS, in which speakers give 5-minute talks.</p><p>The 122nd MTS was held on September 20, 2024.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TxPfiG-dawQ5WYPGpR4hPg.jpeg" /><figcaption>Rob being the MC of the 122nd MTS</figcaption></figure><p><strong>“The Pixel’s Journey” by Simon</strong></p><p>Simon opened the 122nd MTS with his attempt to answer the question, “What does it take to get a pixel from your imagination to your users’ eyeballs?”. The talk goes as far back as the glass teletype and Color Graphics Adapter (CGA). The pixel’s journey back then was considerably simpler than it is today. There is a section of the system memory shared by the CPU and the GPU. The former writes texts (or pixels, for CGA) to it, and the latter renders texts or pixels out of it to the display.</p><p>Rendering 3D scenes is more complicated. CPUs are not fast enough to do it, so GPUs came along to fix the situation. Initially, GPUs work with fixed-function pipelines, which are easy to set up but slow and not very flexible. Programmable pipelines are the opposite; they are difficult to set up but fast. In general, the pixel’s journey within a programmable pipeline goes through a vertex shader (i.e., projecting 3D coordinates to 2D coordinates), a rasterizer (i.e., converting shapes into colorless pixels — “fragments”), and a fragment shader (i.e., coloring the pixels).</p><p>However, the end of the programmable pipeline is not the end of the pixel’s journey. It is only the end of the UI framework. Next comes compositing, a function of modern operating systems that provides applications with an off-screen buffer for each window.</p><p>In summary, the classic pixel’s journey is App → Memory → Display. On the other hand, the modern pixel’s journey is App → UI Framework → Buffer → Composite → Display. The latter consists of more abstraction layers, making timing a bigger challenge. Keeping a consistent timing to prevent screen tearing results in latency. Simon closed his talk with an interesting thought: The latency between server and client kilometers away can be smaller than the latency between client and eyeballs centimeters away.</p><p><strong>“Enhancing Engineering Culture and Inputs Quality With Graphite” by Ryan</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aw-97mjqGOdHDNBEO4MXyg.jpeg" /><figcaption>Ryan presenting “Enhancing Engineering Culture and Inputs Quality With Graphite”</figcaption></figure><p>Output metrics show the result of our work, while input metrics track what we do. Ryan argues that obsessing over output metrics above all else is wrong because we cannot directly control them. Instead, we should focus on improving our input.</p><p>Code quality is the output of a company’s engineering culture. Ryan presented a culture analysis exercise of a hypothetical engineering team: shipping new features takes forever ← code review takes forever ← team members submit large pull requests (PRs) ← PRs are difficult to split into smaller PRs ← smaller PRs are difficult to revise (at least using only Git).</p><p>The two approaches of revising smaller PRs that Ryan explained described the difficulty of the task. The first approach is to stop working. This leads to developers switching to other tasks, which disrupts focus and increases both cognitive load and burnout risk. The second approach is to rebase. This approach has its own challenges as it may lead to merge conflict waves, stale branches, incorrect rebases and tracking overload.</p><p><a href="https://graphite.dev/">Graphite</a> aims to help people create smaller PRs, stay unblocked, and ship faster. Its features include PR inbox, dashboard, GitHub synchronization, and automatic restack. Having used Graphite himself, Ryan mentioned that it allows him to quarantine PRs, identify which PRs cause a particular bug, continue his work off of a particular PR in a team member’s PR stack, and split big PRs after the fact. Ryan also reported that using Graphite led his team to review PRs 50% faster. Going back to his culture analysis exercise, Ryan suggested that this benefit may bubble all the way up to an increase in the speed of shipping new features.</p><p><strong>“A Look Into Huawei’s Programming Language” by Jasmine</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*sQWfu0M19y3wBBLwzl-5FQ.jpeg" /><figcaption>Jasmine presenting “A Look Into Huawei’s Programming Language”</figcaption></figure><p>The <a href="https://cangjie-lang.cn/en/docs">Cangjie</a> programming language was announced by Huawei in June 2024. It is null-safe, statically typed, and has a concurrent garbage collector.</p><p>Jasmine followed a video tutorial to get to know Cangjie. The further she followed the tutorial, the more in touch Jasmine got with her Chinese heritage. This is mainly due to the examples used in the tutorial, some of which relate to calculating cosmic velocities, calculating the zodiac calendar for each month in a particular year, etc.</p><p>Going back to Huawei, Jasmine also came to the realization that many of the company’s products reference Chinese mythology, e.g., <a href="https://www.harmonyos.com/en/">HarmonyOS</a> (referencing the primordial chaotic state of the universe before being split into Heaven and Earth), PanGu LLM (referencing the primordial character related to the creation of the world), and Kunpeng CPUs (referencing a folklore about a giant fish Kun that transformed into a giant bird Peng).</p><p><strong>“HENNGE’s Blueprint for Security” by Jennifer</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*E-cmB0g1iD5mGrXiwzpAmA.jpeg" /><figcaption>Jennifer presenting “HENNGE’s Blueprint for Security”</figcaption></figure><p>In this talk, Jennifer used a house as a metaphor for an information security management system (ISMS). A standard (e.g., <a href="https://www.iso.org/standard/27001">ISO 27001</a>) is to an ISMS what a blueprint is to a house. There are also standard additions (e.g., <a href="https://www.iso.org/standard/43757.html">ISO 27017</a>, <a href="https://www.iso.org/standard/76559.html">ISO 27018</a>, etc.), just as there are additional blueprints for additions to a house. Both an ISMS and a house are also regularly inspected to check that it meets the standard or blueprint requirements, respectively.</p><p>Jennifer explained that ISMS is not the “why” but the “how” of a company’s information security. It is only a framework to be used to provide structure for managing information securely. The “why” comes from the desire to be a company which manages information security properly. If the “why” changes, the “how” may also need to change. Understanding both ISMS and our own work is very important as it allows us to identify conflicts between the rules and the actual situation.</p><p><strong>“Tech to the Rescue: Liberating Language Learning” by Rina-san and Yuta-san</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KdGy4d3pF-AWB22QMYGpww.jpeg" /><figcaption>Rina-san and Yuta-san presenting “Tech to the Rescue: Liberating Language Learning”</figcaption></figure><p>Working full-time and learning a new language is so challenging. Rina-san thought of creating a Slack channel dedicated to promoting learning English. The Slack channel is to be used for posting one English word a day (with its Japanese translation), for posting all relevant announcements written in English, and for its members to freely ask questions about the English language.</p><p>However, after the Slack channel was made, Rina-san found it hard to run the channel manually. It takes time to think of and actually create the contents, and it is easy to forget to post when there are lots of other things to do.</p><p>In response, Yuta-san had the idea to ask Gemini to create the content instead routinely. Gemini retrieves the vocabulary Rina-san is looking for, puts them into Google Sheets, and posts each row of the Google Sheets daily to the Slack channel. Yuta-san then improved this solution by adding the capability of posting interactive messages. The idea is to post a practice test question, then post a hint to its answer 30 minutes later, and finally post the answer 30 more minutes later. Members of the Slack channel answer these questions using emoji reactions. Answers are kept track of to identify difficult questions and post similar practice test questions in the future for further practice.</p><p><strong>“3 Key Takeaways So Far To Enter the US Market” by Nakakomi-san</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VvOCx8OfY99hAF2RWdtULg.jpeg" /><figcaption>Nakakomi-san presenting “3 Key Takeaways So Far To Enter to the US Market”</figcaption></figure><p>Nakakomi-san had a business trip to the US to attend <a href="https://www.saastrannual2024.com/">SaaStr Annual 2024</a>. In this talk, he shared 3 key takeaways from the business trip.</p><p>As always, we had a party afterward 😄</p><p>See you at the next MTS!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=20b99b2e8913" width="1" height="1" alt=""><hr><p><a href="https://medium.com/henngeblog/122nd-monthly-technical-session-20b99b2e8913">122nd Monthly Technical Session</a> was originally published in <a href="https://medium.com/henngeblog">henngeblog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[100 Months of Technical Sessions]]></title>
            <link>https://medium.com/henngeblog/100-months-of-technical-sessions-ebe0676929a6?source=rss-c591be081b7d------2</link>
            <guid isPermaLink="false">https://medium.com/p/ebe0676929a6</guid>
            <category><![CDATA[celebration]]></category>
            <category><![CDATA[technical]]></category>
            <category><![CDATA[hennge]]></category>
            <category><![CDATA[monthly-technical-session]]></category>
            <category><![CDATA[culture]]></category>
            <dc:creator><![CDATA[Bagus Aryabima]]></dc:creator>
            <pubDate>Wed, 25 Jan 2023 02:32:25 GMT</pubDate>
            <atom:updated>2023-01-25T02:32:25.673Z</atom:updated>
            <content:encoded><![CDATA[<p>On November 18, 2022, HENNGE celebrated the 100th installment of Monthly Technical Session (MTS)! On this very special occasion, I represented MTS organizers in presenting a look back to the 8-year history of the event.</p><h3>What is MTS?</h3><p>MTS is a platform for members to share knowledge with their peers. Each speaker gives a talk for 10 - 20 minutes, followed by a Q&amp;A session. Occasionally, there are also lightning talks (5-minute talks, often without a Q&amp;A session afterward) at the end of an MTS.</p><h3>1st Era: May 2014 - September 2016</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*P-LCBQTjmJElfK6-7n23iQ.jpeg" /><figcaption>Maki-san presenting <strong>Fighting With JSON Schema and Scaffolding Go Apps</strong> in MTS #20 (March 18, 2016)</figcaption></figure><p>In the 1st Era, MTSs were held on 10F.</p><p>MTS #1 was held on May 30, 2014. Unfortunately, I haven’t been able to find any photos or videos of this inaugural event.</p><p>I did find a video of MTS #3 (July 25, 2014), probably the first MTS to be recorded on video.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*hCikNGb-Ep6irF9Y" /><figcaption>Iskandar presenting <strong>Automated GIP Admission System</strong> in MTS #3 (July 25, 2014)</figcaption></figure><p>MTS #6 (October 24, 2014) is the first MTS to be recorded in <a href="https://blog.hde.co.jp/entry/2014/11/27/203626">a blog post on HDE Blog</a>. HDE Blog is the precursor to HENNGE Blog, the Medium publication you are currently reading.</p><p>The 1st Era includes the only months in which no MTS was held, i.e.</p><ul><li>December 2014</li><li>August 2015</li><li>November 2015</li></ul><h3>2nd Era: October 2016 - September 2019</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fgLzZPp2UOln5xdAmLRZRA.jpeg" /><figcaption>Beer Bash of MTS #36 (July 21, 2017)</figcaption></figure><p>In the 2nd Era, MTSs were held on 5F. This is the longest Era so far.</p><p>MTS #27 (October 21, 2016) is the first MTS to be recorded in <a href="https://blog.hde.co.jp/entry/2016/10/27/155748">a blog post written in English</a>, still on HDE Blog. MTS #44 (March 23, 2018) is the first MTS to be recorded in <a href="https://medium.com/henngeblog/hde-tech-44th-monthly-technical-session-mts-report-d7d666a7962a">a blog post on HENNGE Blog</a>.</p><p>On September 21, 2018, HENNGE (or rather HDE, i.e., the company name at the time) celebrated the 50th installment of MTS!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aVPBQAn7peKto0GT0D0bbA.jpeg" /><figcaption>🎉 MTS #50 (September 21, 2018) Celebration Cake 🎉</figcaption></figure><h3>3rd Era: October 2019 - February 2020</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pKCCeJoMy9arvmyw7Vxv2A.jpeg" /><figcaption>GIP Honor Day in MTS #67 (February 28, 2020)</figcaption></figure><p>In the 3rd Era, MTSs were held on 4F. This is the shortest Era so far.</p><p>You probably already figured out the reason why.</p><h3>4th Era: March 2020 - April 2022</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-nGweam9lFvyIvGVfaEOzQ.png" /><figcaption>Tanabe-san presenting <strong>Cgo Is Not Go</strong> in MTS #68 (March 27, 2020)</figcaption></figure><p>Yup, COVID-19 happened. So in the 4th Era, MTSs were held on Zoom.</p><p>For the first few months of the 4th Era, MTS organizers tried to hold MTSs both on 4F and on Zoom, i.e., hybrid MTSs.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Ey-0MxLQdDvIjXcgyCgtlg.png" /><figcaption>GIP Honor Day in MTS #69 (April 24, 2020) was held on 4F and streamed on Zoom</figcaption></figure><p>However, by the time MTS #73 (August 21, 2020) came around, it had been decided to keep holding fully-remote MTSs, due to the increasing number of COVID-19 cases and the complexity of holding hybrid MTSs.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WY29OXQO-GZlcU98SYE1SQ.png" /><figcaption>Beer Bash of MTS #77 (December 18, 2020)</figcaption></figure><h3>5th Era: May 2022 - Today</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*9lEWE0sW8mI7M4VgU8Y-qQ.jpeg" /><figcaption>Hiroon presenting <strong>Microsoft Power Platform: The Future of App-Building</strong> in MTS #96 (July 22, 2022)</figcaption></figure><p>In the 5th Era, MTS is held on both 4F and Zoom. This is going to be <em>the</em> setup for MTSs in the foreseeable future.</p><p>By the 5th Era, the number of COVID-19 cases had decreased and MTS organizers had figured out an approach of holding hybrid MTSs. Today, MTS organizers are experienced enough in holding hybrid events that they share their knowledge with organizers of other events in HENNGE.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gRsPbxUuJmZZjpWt6y4OXg.jpeg" /><figcaption>Beer Bash of MTS #98 (September 16, 2022)</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_vJkxFhg3TEwiw9p_B_asA.png" /><figcaption>The first 100 months of technical sessions</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ebe0676929a6" width="1" height="1" alt=""><hr><p><a href="https://medium.com/henngeblog/100-months-of-technical-sessions-ebe0676929a6">100 Months of Technical Sessions</a> was originally published in <a href="https://medium.com/henngeblog">henngeblog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[92nd Monthly Technical Session]]></title>
            <link>https://medium.com/henngeblog/92nd-monthly-technical-session-fe8ade3eaf6f?source=rss-c591be081b7d------2</link>
            <guid isPermaLink="false">https://medium.com/p/fe8ade3eaf6f</guid>
            <category><![CDATA[hennge]]></category>
            <category><![CDATA[pulumi]]></category>
            <category><![CDATA[ngrok]]></category>
            <category><![CDATA[monthly-technical-session]]></category>
            <category><![CDATA[gpu-programming]]></category>
            <dc:creator><![CDATA[Bagus Aryabima]]></dc:creator>
            <pubDate>Tue, 31 May 2022 03:02:55 GMT</pubDate>
            <atom:updated>2022-05-31T03:02:55.114Z</atom:updated>
            <content:encoded><![CDATA[<p>HENNGE’s Monthly Technical Session (MTS) is a platform for members to share knowledge with their peers. Each speaker gives a talk for 10–20 minutes, followed by a Q&amp;A session. Occasionally there is also a lightning talk session at the end of an MTS, in which speakers give 5-minute talks.</p><p>The 92nd MTS was held on March 25, 2022, on Zoom. This time we also had some talks with our <a href="https://hennge.com/global/gip.html">Global Internship Program</a> (GIP) interns.</p><h3>1. “Introduction to ngrok” by Kodama-san</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*p1fHFHWIuWpKmsNJ4Thp3Q.png" /><figcaption>Kodama-san presenting “Introduction to ngrok”</figcaption></figure><p>Kodama-san opened the 92nd MTS with his talk about the basics of <a href="https://ngrok.com/">ngrok</a>, a globally distributed reverse proxy fronting web service running in any cloud or private network (or even our own machines). With ngrok, we can expose web servers running on our machines to the Internet. ngrok provides public URLs for testing on mobile devices, testing chatbots, sending previews to clients, SSH access to Raspberry Pis, building webhook integrations, etc.</p><p>Like most SaaS, ngrok has several pricing plans, including a free one. The Free plan is perhaps most appropriate for testing purposes. On the other hand, the other plans include advanced features such as automatic generation of SSL certificates for custom domains, authentication delegation, management API, event streaming, native OS service management, load balancing, request routing, and TLS client authentication. Kodama-san and his teammates are researching how we can use those advanced features to improve HENNGE’s products and services.</p><h3>2. “Service Mesh” by Nikhil</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*IAr8VZMEd2SemL_xnJqMNA.png" /><figcaption>Nikhil presenting “Service Mesh”</figcaption></figure><p>Nikhil, one of our GIP interns, gave a talk about service mesh, a dedicated infrastructure layer for facilitating service-to-service communications between services or microservices, using a proxy. The focus of Nikhil’s talk is on how service meshes can help us overcome the challenges of microservices architecture.</p><p>Microservices architecture (also known simply as microservices) is an architectural style for developing applications. It allows a large application to be separated into smaller independent parts, with each part having its own realm of responsibility.</p><p>According to Nikhil, here are some of the benefits of microservices:</p><ul><li>Faster delivery: microservices allow frequent releases with less risk</li><li>Isolation: it is unlikely for a single service to crash an entire system</li><li>Scaling: each service can be scaled depending on their usage</li><li>Culture: each team works on a single business capability</li></ul><p>At the same time, microservices present certain challenges:</p><ul><li>Service discovery: automatically detecting services on a network</li><li>Traffic splitting: specifying the distribution of traffic across multiple versions of a service (e.g. for deployments, A/B testing, service migration)</li><li>Security: instead of a single entry point, microservices offer many, with each potentially being a point of vulnerability</li><li>Monitoring: understanding the behavior of a microservices application and troubleshooting its problems are challenging since requests span multiple services, with each performing its own operations</li></ul><p>Service mesh promises to be a solution to those challenges. A service mesh consists of services and proxies running as sidecars to the services. A sidecar is attached to a parent service and provides supporting features for the parent service. A service mesh combines its sidecars into a proper distributed system, including a data plane and a control plane. The former manages sidecars, while the latter performs service discovery, service routing, load balancing, authentication, and authorization, among others.</p><h3>3. “GPU Programming” by Steve</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_kiNt-8jvmvVnM2jjIBN5Q.png" /><figcaption>Steve presenting “GPU Programming”</figcaption></figure><p>Steve, another one of our GIP interns, gave a talk about GPU programming. GPUs were invented out of the need to render 3D images to the screen (i.e. pixels) using the graphics pipeline, which is basically a number of matrix transformations. As the pipeline grew more complex, CPU processing could no longer keep up with it. And thus the GPU was invented to handle tasks like the graphics pipeline (i.e. single instruction, multiple data — SIMD). Today GPUs are also used in machine learning, crypto mining, image processing, and other SIMD tasks.</p><p>Compared to CPUs, GPUs have smaller but many, <em>many</em> more cores, most of which are arithmetic-logic units (ALUs), control units, and caches. This is intentional, as CPUs are designed for serialized executions, low latency, and more flexibility. On the other hand, GPUs are designed for parallelized processing, high throughput, and fewer design controls.</p><p>So, how do we program GPUs? There are many GPU programming frameworks out there, such as <a href="https://developer.nvidia.com/cuda-toolkit">NVIDIA CUDA</a>, <a href="https://www.khronos.org/opencl/">OpenCL</a>, and <a href="https://www.openacc.org/">OpenACC</a>. The most popular out of these is CUDA, to the point that it is also used internally by machine learning and deep learning frameworks like <a href="https://www.tensorflow.org/">Tensorflow</a> and <a href="https://pytorch.org/">PyTorch</a>.</p><p>CUDA is a toolkit that extends C and C++ by allowing us to define functions called kernels. When a kernel is called, it is executed N times in parallel by N different CUDA threads (as opposed to only once in the case of regular functions). Each thread that executes the kernel runs independently and has access to various useful information, such as its own thread index, through built-in variables.</p><p>Steve compared the performances of a CPU, a single CUDA thread, and 256 CUDA threads in adding 2 arrays, each containing a million elements:</p><ul><li>CPU: ~70ms</li><li>1 CUDA thread: ~400ms</li><li>256 CUDA threads: ~2ms</li></ul><p>This is an example of a SIMD task, and considering the number of threads used, the 256 CUDA threads came out on top as expected.</p><h3>4. “Go 1.18 and Fuzzing” by Tanabe</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KSTjeluuukt1zymRK53EQA.png" /><figcaption>Tanabe-san presenting “Go 1.18 and Fuzzing”</figcaption></figure><p><a href="https://go.dev/blog/go1.18">Go 1.18</a> was released on March 15, 2022. Fuzzing is one of the new features included in this version. It is the act of automatically seeding random data against tests in an attempt to find vulnerabilities or crash-causing inputs.</p><p>Fuzzing is recommended for programs that are expected to be robust even when given <em>any</em> kind of input, as it is really hard to manually produce certain inputs that hit edge cases. For example, <a href="https://www.chromium.org/Home/">Chromium</a> is continuously fuzzed by the Chrome Security Team, utilizing 15,000 cores (Mozilla and Microsoft also fuzz their browsers). The Heartbleed vulnerability was also <a href="https://forallsecure.com/blog/the-fuzzing-files-the-anatomy-of-a-heartbleed">discovered by fuzzing OpenSSL</a>.</p><p>Fuzzing is integrated with the standard toolchain of the “go test” command. However, since fuzzing is <em>very</em> resource-intensive, it has to be invoked manually by using the “-fuzz” flag. Failing inputs are automatically saved to the “testdata/fuzz/&lt;TestName&gt;/&lt;Input&gt;” directory. Fuzzing is done concurrently using multiple CPU cores, and it indeed uses a lot of CPU capacity (as well as storage). As of Go 1.18, coverage-guided fuzzing is available for AMD64 and ARM64 architectures as a more efficient alternative.</p><p>Tanabe-san and his teammates had tried fuzzing their products and services. After several hours, a crash bug was found and the team then worked on fixing it.</p><p>At the end of his talk, Tanabe-san gave some tips on writing fuzzing tests:</p><ul><li>Follow the standard library’ convention in naming fuzzing tests (i.e. &lt;TestName&gt;_test.go)</li><li>Write a round-trip fuzzing test whenever possible (i.e. parse → encode → parse → compare the result; for successful parses, assert that parsers are in the “known” state)</li><li>Start writing fuzzing tests so that as many errors as possible fail early, then update the code to ignore known errors and carry on fuzzing (make use of the fact that “go test” automatically records all failing inputs)</li></ul><h3>5. “Infrastructure as Code with Pulumi” by Krizza</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Hk5nGeCSHwleugK5UbD1Gg.png" /><figcaption>Krizza presenting “Infrastructure as Code with Pulumi”</figcaption></figure><p>Krizza was the last of our GIP interns to give a talk on the 92nd MTS. She introduced us to <a href="https://www.pulumi.com/">Pulumi</a>, a so-called developer-first Infrastructure as Code (IaC).</p><p>These days there is no shortage of IaC tools, such as <a href="https://aws.amazon.com/cloudformation/">AWS CloudFormation</a>, <a href="https://www.terraform.io/">Terraform</a>, and Pulumi, among others. AWS CloudFormation is exclusively for AWS services, written in JSON or YAML, and — according to Krizza — doesn’t provide the best developer experience. Terraform, on the other hand, is cloud-agnostic and written in HCL. Pulumi is quite unique as it is built on top of existing tech (i.e. other programming languages) and is open source. With Pulumi, we can provision infrastructure using familiar languages, such as TypeScript, JavaScript, Python, Go, C#, etc.</p><p>Krizza spent most of her talk actually demonstrating infrastructure provisioning with Pulumi. The programming language she chose for this demo was Python.</p><h3>6. Lightning Talk #1: “Criminal Actor Targeting Organizations for Data Exfiltration and Destruction” by Ogura-san</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CFaPJvggJeK8FBZZDktkIQ.png" /><figcaption>Ogura-san presenting “Criminal Actor Targeting Organizations for Data Exfiltration and Destruction”</figcaption></figure><p>In relation to <a href="https://www.wsj.com/articles/okta-investigates-reports-of-a-digital-breach-11647931174">Okta’s investigation of digital breach reports</a>, Microsoft has released <a href="https://www.microsoft.com/security/blog/2022/03/22/dev-0537-criminal-actor-targeting-organizations-for-data-exfiltration-and-destruction/">a report</a> detailing their own investigation into large-scale social engineering and extortion campaigns against multiple organizations (one of which being Okta). Ogura-san used his lightning talk to raise awareness of both the issue and the report and encouraged everyone to think about what we can do to protect ourselves from such campaigns.</p><h3>7. Lightning Talk #2: “How to Write For-Loops Without Actually Writing For-Loops (in Python)” by Jonas</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*prHKjl3s49vg-I5zl85E-Q.png" /><figcaption>Jonas presenting “How to Write For-Loops Without Actually Writing For-Loops (in Python)”</figcaption></figure><p>No, Jonas’ talk was not about recursion.</p><p>Instead, he used a combination of decorators, unicodes in identifiers, regexes, and ASCII codes to replace actual code objects. The idea is that the code object of the decorated function is replaced with an updated version of itself (the update being replacing certain identifiers with ASCII codes, which builds the strings of for-loops 😁).</p><p>As usual, we had a party afterward (on Zoom) 😄</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ji_cC5SJjkDHU5eBlVykuw.png" /><figcaption>92nd MTS after-party</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fe8ade3eaf6f" width="1" height="1" alt=""><hr><p><a href="https://medium.com/henngeblog/92nd-monthly-technical-session-fe8ade3eaf6f">92nd Monthly Technical Session</a> was originally published in <a href="https://medium.com/henngeblog">henngeblog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[65th Monthly Technical Session]]></title>
            <link>https://medium.com/henngeblog/65th-monthly-technical-session-6c303386838a?source=rss-c591be081b7d------2</link>
            <guid isPermaLink="false">https://medium.com/p/6c303386838a</guid>
            <category><![CDATA[hennge]]></category>
            <category><![CDATA[monthly-technical-session]]></category>
            <category><![CDATA[japan]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[Bagus Aryabima]]></dc:creator>
            <pubDate>Tue, 25 Feb 2020 02:31:01 GMT</pubDate>
            <atom:updated>2020-02-25T02:31:01.350Z</atom:updated>
            <content:encoded><![CDATA[<p>Monthly Technical Session (MTS) is HENNGE’s mini-conference. As the name implies, MTS is held monthly and its talks are (mostly) about technology.</p><p>The 65th MTS was held on December 20th, 2019.</p><h3>“About DMARC” by Yagmur</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*AlH9DX2REOu15B-ujxibYQ.jpeg" /><figcaption>Yagmur presenting “About DMARC”</figcaption></figure><p>Domain-based Message Authentication Reporting &amp; Conformance (DMARC) is an email authentication, policy, and reporting protocol. It helps protecting email senders and recipients from spam, spoofing, and phishing.</p><p>DMARC builds on Sender Policy Framework (SPF) and DomainKeys Identified Mail (DKIM) protocols. SPF verifies whether an email message was sent by an authorized sender. DKIM verifies that the contents of an email message haven’t been tampered with. Not only does DMARC require that email messages are protected by both SPF and DKIM, it also requires the domains used by SPF or DKIM to align with the domain found in the “From” address.</p><p>By specifying DMARC policies, it is possible to determine how to handle email messages that do not pass DMARC validation. Report-only mode accepts such email messages, quarantine mode quarantines them (e.g. sends them to the spam folder), and reject mode makes it so that such email messages do not reach the end user.</p><h3>“A Gentle Introduction to Cross-Platform App Development” by Kelvin</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rIm57flLZQZKuAvqVg8UvA.jpeg" /><figcaption>Kelvin presenting “A Gentle Introduction to Cross-Platform App Development”</figcaption></figure><p>Kelvin explained that there are several merits that can be gained from developing cross-platform apps: code reuse, single codebase, unified UI and logic, and (possibly) smaller number of developers. There are several frameworks we can use in developing cross-platform apps. Kelvin shared his thoughts about some of these frameworks.</p><p><a href="https://phonegap.com/">PhoneGap</a> and <a href="https://cordova.apache.org/">Cordova</a> is essentially a responsive website running on a local server with additional privileges. Using these frameworks enables web developers to develop cross-platform mobile apps. Unfortunately, apps developed using these frameworks tend to not look so good.</p><p>Most of the time, <a href="https://dotnet.microsoft.com/apps/xamarin">Xamarin</a> is used by development teams that specializes in .NET framework. Xamarin offers platform-specific UI implementations, but this feature diminishes the code reuse benefit somewhat.</p><p><a href="https://facebook.github.io/react-native/">React Native</a> is one of the most prominent cross-platform app development framework these days. Similar to PhoneGap and Cordova, this framework is well-suited for web developers, especially ones experienced in using <a href="https://reactjs.org/">React</a>. React Native doesn’t exactly map 1-to-1 to React however, and its UI output relies on its bridge concept.</p><p><a href="https://flutter.dev/">Flutter</a> works like a cross-platform game engine (e.g. <a href="https://unity.com/">Unity</a>, <a href="https://www.unrealengine.com/en-US/">Unreal Engine</a>) in that it draws its own pixels. It makes use of the <a href="https://dart.dev/">Dart</a> programming language, which according to Kelvin is a plus. However, it is indeed still very young. It is also backed by Google, which implies there are precedents for Flutter to be rather short-lived :(</p><p>Finally, Kelvin mentioned some caveats to the aforementioned benefits of cross-platform development. Depending on the framework, there may not be much code to be reused (e.g. Xamarin) and there may be a need for <em>more</em> developers (e.g. specialists in cross-platform framework languages).</p><h3>“Flutter” by Yazeed</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QHwOWvstHAyk8Nq01VqaQA.jpeg" /><figcaption>Yazeed presenting “Flutter”</figcaption></figure><p>The original iPhone was released in 2007. The first commercial Android device was launched the following year. Today, there are <em>much more</em> devices, screen sizes, and OS versions that mobile app developers have to work with. Cross-platform app development frameworks such as Flutter mean to ease this pain.</p><p>The UI of Flutter apps is made of widgets. Yazeed likened them to Lego blocks of mobile applications. He also demonstrated how they work by coding a simple web application live. During the live coding session, Yazeed showcased features of both the Flutter framework and the Dart programming language.</p><h3>“re:Invent 2019” by Bumi, Doi, and Michelle</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rDVVeuR4cVjEnGyUPqSUJA.jpeg" /><figcaption>Michelle presenting “re:Invent 2019”</figcaption></figure><p>Bumi, Doi, Michelle and other HENNGE members had gone to and returned from Las Vegas to attend re:Invent 2019, a learning conference hosted by Amazon Web Services (AWS). The event featured keynote announcements, training and certification opportunities, more than 2,500 technical sessions, and a partner expo, among other things.</p><p>Michelle had been particularly interested in technical sessions that covered machine learning (ML) and artificial intelligence (AI). AWS offers many solutions for various ML and AI needs, on both the AI service level and ML infrastructure level. Michelle learned how to develop an Alexa skill that reminds her of her birthday, got to know more about ML and AI components of Amplify, and composed some music on the DeepComposer workshop.</p><p>Bumi talked about DeepComposer in more detail. Like DeepLens and DeepRacer before it, DeepComposer is yet another effort by AWS in putting ML in the hands of every developer. Each of these services teaches us different lessons about ML. In the case of DeepComposer, the lesson is about using generative adversarial network (GAN) to compose music. The way it works seems simple enough: input melody, then choose genre or make a new model to generate music. We can then publish the music to SoundCloud or edit them in our preferred digital audio workstations (DAW). Behind the scenes, DeepComposer makes use of other AWS services such as Amazon SageMaker and AWS Lambda.</p><p>Finally, Doi-san shared his experience of having a meeting with one of AWS’ product teams during re:Invent. Contrary to his expectations, this meeting is <em>not</em> about giving suggestions to these product teams. What ended up happening was more akin to a consultation, in that the product teams gave Doi-san some advice on solving his problems.</p><h3>“What Web Summit 2019 Has Taught Us” by Jazz and Kameyama</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*N3dVmNzM3ARBOpYdY6i_hA.jpeg" /><figcaption>Jazz (left) and Kameyama (right) presenting “What WebSummit 2019 Has Taught Us”</figcaption></figure><p>Jazz, Kameyama-san, and several other HENNGE members had had the chance to attend Web Summit 2019. This annual technology conference gathers the founders and CEOs of technology companies, fast-growing startups, policymakers even heads of state to discuss the global tech industry. There were more than 70,000 attendees this year from more than 160 countries. Web Summit 2019 offered 23 tracks, in which more than 1,200 speakers covered topics such as autonomous vehicles, financial technology, and AI and robotics.</p><p>Jazz and Kameyama-san had learned a lot from this event. From a company that recently successfully rebranded themselves, Mailchimp, they learned how they communicate their identity through the touchpoints with their customers. Every touch points you have with your customers are part of your brand, e.g. logo, color, website, SNS, font, etc. These touchpoints need to be designed in a way that effectively communicates the emotions you want your customers to feel.</p><p>The day of the 65th MTS was also the last day of our Global Internship Program (<a href="https://www.hennge.com/global/gip/">GIP</a>) intern, Yazeed. We held a small event for him and gave him some souvenirs. In turn, he shared his impressions of working with us.</p><p>Thank you very much, Yazeed!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*w8Nuoon0I1lc1IGuyVy8AA.jpeg" /><figcaption>Yazeed holding his GIP certificate</figcaption></figure><p>As usual, we had a party afterwards :)</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HHqtOF45tXhYiCz2rYkFTg.jpeg" /><figcaption>65th MTS after-party</figcaption></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6c303386838a" width="1" height="1" alt=""><hr><p><a href="https://medium.com/henngeblog/65th-monthly-technical-session-6c303386838a">65th Monthly Technical Session</a> was originally published in <a href="https://medium.com/henngeblog">henngeblog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Day 19 — Indonesian Restaurants in Tokyo]]></title>
            <link>https://medium.com/henngeblog/day-19-indonesian-restaurants-in-tokyo-3621bf883edf?source=rss-c591be081b7d------2</link>
            <guid isPermaLink="false">https://medium.com/p/3621bf883edf</guid>
            <category><![CDATA[henngeadvent2019]]></category>
            <category><![CDATA[indonesia]]></category>
            <category><![CDATA[henngeblog]]></category>
            <category><![CDATA[food]]></category>
            <dc:creator><![CDATA[Bagus Aryabima]]></dc:creator>
            <pubDate>Thu, 19 Dec 2019 04:01:02 GMT</pubDate>
            <atom:updated>2019-12-19T04:01:02.532Z</atom:updated>
            <content:encoded><![CDATA[<h3>Day 19 — Indonesian Restaurants in Tokyo</h3><p>Hello everyone, my name is Bagus. I am a software engineer from Indonesia.</p><p>I wrote <a href="http://hde-advent-2015.hatenadiary.jp/entry/2015/12/09/130317">my first HENNGE Advent Calendar article</a> in 2015. Through the article, I tried to introduce the food of my people to the world.</p><p>During the last 4 years, I have eaten at Indonesian restaurants in Tokyo several times. I would like to share with you what I think about some of them.</p><h3>Ayung Teras</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*ArwOADm822qBeg7O" /><figcaption>Photo by <a href="https://tabelog.com/rvwr/003297907/">y_rec</a> on <a href="https://tabelog.com/en/tokyo/A1303/A130301/13004641/">Tabelog</a></figcaption></figure><p><strong>Nearest station</strong>: Shibuya</p><p>Ayung Teras is the closest Indonesian restaurant from HENNGE Shibuya HQ. Me and other members of HENNGE who come from Indonesia have eaten here many, many times over the years. Some members also like to bring our <a href="https://hennge.com/global/gip/">GIP interns</a> here for lunches.</p><p>To be honest, the taste of Ayung Teras’ foods is fine. There was a time when I thought that its location was a stronger reason to return there to eat, rather than the taste of its food. I changed my mind after having dinner there for the first time. I ate <em>soto</em> (Indonesian soup), <em>nasi goreng</em> (fried rice), and a couple other Indonesian foods that night. Honestly, they were all delicious!</p><p>Having said that, I do think that Ayung Teras’ foods do not taste as strong as authentic Indonesian foods. For people who are interested in eating Indonesian food but never have had the chance to, I recommend you to start the process at Ayung Teras. This will ease you into enjoying flavorful Indonesian foods :)</p><h3>Cabe</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*sAoxTv_mHxna5Vve" /><figcaption>Photo by <a href="https://tabelog.com/rvwr/001801775/">YamaNe79</a> on <a href="https://tabelog.com/en/tokyo/A1316/A131601/13027513/">Tabelog</a></figcaption></figure><p><strong>Nearest station</strong>: Meguro</p><p><strong>Fun fact</strong>: Cabe means chili pepper in Indonesian :)</p><p>Cabe is arguably my favorite Indonesian restaurant in Tokyo. There are so many foods on the menu, most — if not all of them — tastes authentic <em>and</em> delicious. Consider having <em>nasi campur</em> (mixed rice) for lunch if you visit during the day. Cabe’s take on this dish consists of white rice, fried noodles, Indonesian pickles, Indonesian curry, fritters, and fried chicken. You’ll get to eat many kinds of food for a reasonable price! Of course, you can’t go wrong with foods such as <em>rendang</em> (spicy meat dish) or <em>sate</em> (skewered meat) either.</p><p>Not only does Cabe has good appetizers and main courses, but it also has a selection of traditional Indonesian desserts on the menu. They offer <em>bubur ketan hitam</em> (black gelatinous rice porridge), <em>kolak</em> (sweet potato in coconut milk base), and <em>pisang goreng</em> (fried banana), among other things. I am not too fond of these desserts in general, but I tried Cabe’s deserts and can certainly vouch for their authenticity.</p><p>If you’re looking for food with stronger tastes, or if you are ready to ‘graduate’ from Ayung Teras, I do recommend you to visit Cabe. They are unfortunately closed on Sundays however, so your best bet would be to come here on Saturdays.</p><h3>Dapoer Koneng</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tX3Rd8s6omWJAYBWU1LfNg.png" /><figcaption>Photo by Darang Melati Z on <a href="https://www.google.com/maps/place/Dapoer+Koneng+Indonesian+Cafe/@35.474904,139.626532,3a,80y,90t/data=!3m8!1e2!3m6!1sAF1QipNO696fMZrPvNFtTRNrCFV5VBFjvUmldoG1tFY1!2e10!3e12!6shttps:%2F%2Flh5.googleusercontent.com%2Fp%2FAF1QipNO696fMZrPvNFtTRNrCFV5VBFjvUmldoG1tFY1%3Dw152-h86-k-no!7i1280!8i720!4m12!1m6!3m5!1s0x60185c19c77a6163:0x48fd0397d31f89f!2sDapoer+Koneng+Indonesian+Cafe!8m2!3d35.474904!4d139.626532!3m4!1s0x60185c19c77a6163:0x48fd0397d31f89f!8m2!3d35.474904!4d139.626532">Google Maps</a></figcaption></figure><p><strong>Nearest station</strong>: Tammachi</p><p><strong>Fun fact</strong>: Dapoer Koneng means Yellow Kitchen in Sundanese :)</p><p>Dapoer Koneng makes for a strong case of my favorite Indonesian restaurant in Tokyo. It is quite similar to Cabe in terms of taste. Naturally then, one of the distinguishing factors between the two is the menu. Dapoer Koneng offers some food that Cabe doesn’t and vice versa. For example, I ate <em>pempek</em> (fishcake) and <em>rawon</em> (beef black soup) in Dapoer Koneng, and I have yet to see those dishes in Cabe if I remember correctly.</p><p>The same applies to their dessert offerings. Personally, Dapoer Koneng wins in this department, because it has my favorite Indonesian desert: <em>es campur</em> (mixed shaved ice with fruits, condensed milk, and syrup).</p><p>Dapoer Koneng has something that perhaps no other Indonesian restaurants have: Indonesian karaoke machine. Yes, there is a karaoke machine in this restaurant, and you can sing Indonesian songs with it. On my second visit to Dapoer Koneng, I saw some people who were singing their hearts out :)</p><p>So there you go, those were my thoughts about 3 Indonesian restaurants in Tokyo. I have been to other Indonesian restaurants before, but I visited these 3 restaurants the most. There are also some Indonesian restaurants to which I have never gone so far, such as <strong>Cinta Jawa</strong> in Akihabara and <strong>Monggo Moro</strong> in Shinjuku. I have been hearing good things about these places, so I can’t wait to visit them. Perhaps I’ll write what I think about them in HENNGE Advent Calendar 2020!</p><p>Thank you for reading. I hope this article piqued your interest in enjoying Indonesian cuisine.</p><p><em>This article is part of the HENNGE Advent Calendar 2019. An Advent calendar is a special calendar used for counting down the days till Christmas. HENNGE Advent Calendar 2019 presents one article by one HENNGE member per day for 25 days until Christmas, 2019</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3621bf883edf" width="1" height="1" alt=""><hr><p><a href="https://medium.com/henngeblog/day-19-indonesian-restaurants-in-tokyo-3621bf883edf">Day 19 — Indonesian Restaurants in Tokyo</a> was originally published in <a href="https://medium.com/henngeblog">henngeblog</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>