<?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 Muhammad Azmi on Medium]]></title>
        <description><![CDATA[Stories by Muhammad Azmi on Medium]]></description>
        <link>https://medium.com/@mhmmdazmye?source=rss-05407ce5c0ef------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*4yKkN6t7JWRVfbNyEpwq1Q.jpeg</url>
            <title>Stories by Muhammad Azmi on Medium</title>
            <link>https://medium.com/@mhmmdazmye?source=rss-05407ce5c0ef------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 21:01:27 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@mhmmdazmye/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[How Pre-AI Engineers Actually Use AI: Architecture First, Prompts Second]]></title>
            <link>https://medium.com/@mhmmdazmye/how-pre-ai-engineers-actually-use-ai-architecture-first-prompts-second-af9daff402d2?source=rss-05407ce5c0ef------2</link>
            <guid isPermaLink="false">https://medium.com/p/af9daff402d2</guid>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[full-stack-developer]]></category>
            <category><![CDATA[software-engineering]]></category>
            <dc:creator><![CDATA[Muhammad Azmi]]></dc:creator>
            <pubDate>Wed, 29 Jul 2026 18:33:42 GMT</pubDate>
            <atom:updated>2026-07-29T18:33:42.788Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Q5tw74Vl3sThuPI2SLanqw.png" /><figcaption>AI Generated Image</figcaption></figure><h4>Why senior developers don’t ask AI to “build this with zero mistakes” — and how experienced engineers use LLMs to craft clean, maintainable, and scalable software.</h4><p>If you listen to the current hype around generative AI, you might believe that software engineering has been reduced to a single prompt:</p><blockquote>“AI, build me an enterprise web app with zero bugs, clean code, and perfect UI.”</blockquote><p>To anyone who has spent years in the trenches — building complex Electronic Medical Record (EMR) systems for public hospitals , high-stakes ERP platforms , or high-performance frontend landing pages — this narrative sounds absurd. Software engineering has never been just about writing syntax. Syntax is merely the final step of a long series of structural, architectural, and business decisions.</p><p>Engineers who learned to build software before the AI boom aren’t resisting AI; they are adapting to it. But their workflow looks fundamentally different from that of a beginner “vibe coder.”</p><p>An experienced developer doesn’t use AI as a magic machine to spit out whole applications from thin air. Instead, they treat AI as an elite junior developer or co-pilot — one that requires strong architectural boundaries, clear conventions, and disciplined supervision to produce clean, maintainable code.</p><p>Here is how engineers who understand software foundations adapt and thrive in the AI era.</p><h3>1. The Trap of “AI, Build Me This With Zero Mistakes”</h3><p>When you ask an AI to generate an entire project from scratch without guardrails, it will usually give you something that <em>looks</em> working on the surface. But look under the hood, and you’ll often find a ticking time bomb:</p><ul><li><strong>Inconsistent Folder Structures:</strong> Components scattered randomly, mixing business logic directly into presentation views.</li><li><strong>Tightly Coupled State:</strong> State management paradigms changing from one file to the next.</li><li><strong>Hidden Tech Debt:</strong> AI choosing quick hacks over scalable design patterns, creating duplication and high cognitive complexity.</li></ul><p>AI model contexts are stateless by default. They don’t know your long-term roadmap, your team’s code standards, or how your system will handle scale unless you explicitly enforce that structure. Expecting AI to architect a system perfectly without human direction is a fast track to spaghetti code.</p><h3>2. Step 1: Human Initialization &amp; System Architecture</h3><p>Before firing up a single prompt or opening an AI coding assistant, an experienced engineer does what they have always done: <strong>they build the foundation.</strong></p><ol><li><strong>Initializing the Repository:</strong> Manually bootstrapping the project using modern, lightweight tooling (like Vite, React, Next.js, or Hono.js).</li><li><strong>Structuring the Architecture:</strong> Defining clear directory boundaries — whether that’s Feature-Sliced Design (FSD), Domain-Driven Design (DDD), or a clean modular folder layout.</li><li><strong>Establishing Core Patterns:</strong> Setting up centralized state management, API request layers, error boundaries, and environment handling.</li></ol><p>By setting up the scaffold manually, you establish the “laws of physics” for your codebase. When the AI eventually operates within this repository, it reads these existing patterns and stays within the established boundaries instead of hallucinating its own architecture.</p><h3>3. Step 2: Defining the Code Writing Style &amp; Guardrails</h3><p>An AI model’s output quality is directly constrained by the context you feed it. Experienced engineers don’t rely on generic AI defaults; they explicitly define their project’s coding standards upfront.</p><p>This includes setting up:</p><ul><li><strong>Strict Linter &amp; Formatting Rules:</strong> ESLint, Prettier, and strict TypeScript configurations.</li><li><strong>Coding Style Guidelines:</strong> Specifying explicit rules (e.g., <em>“Use functional components, composition over inheritance, custom hooks for logic separation, and strict atomic CSS classes”</em>).</li><li><strong>AI Context Rules (</strong><strong>.cursorrules or System Instructions):</strong> Explicitly telling the AI engine how code should be structured in this repository before asking it to implement a feature.</li></ul><p>When you pre-define the rules of engagement, the AI isn’t guessing how to write code — it is mirroring your established standards.</p><h3>4. Step 3: Deploying AI as a Refinement &amp; Speed Engine</h3><p>Once the project is initialized, the architecture is clear, and the code style is enforced, AI becomes an extraordinary multiplier.</p><p>Instead of asking AI to “build the app,” pre-AI engineers break down execution into targeted, high-value tasks:</p><h4>A. Accelerating Boilerplate Execution</h4><p>Need to write 15 typed TypeScript interfaces from a backend API response schema? Or build standard CRUD handlers following your existing repository pattern? This is where AI excels — doing the repetitive typing in seconds so the engineer can focus on core logic.</p><h4>B. Enforcing Clean &amp; Maintainable Code</h4><p>Experienced engineers use AI to refactor and optimize code that has already been drafted.</p><blockquote>“Refactor this custom React hook to reduce cognitive complexity, ensure proper cleanup in useEffect, and follow our modular error-handling pattern.”</blockquote><h4>C. Finding Edge Cases &amp; Security Flaws</h4><p>Instead of trusting AI blindly, seasoned engineers turn the tables and ask AI to audit human logic:</p><blockquote>“Review this form validation layer. What edge cases or memory leaks am I missing when handling concurrent user inputs?”</blockquote><h4>D. Automated Unit Testing &amp; Documentation</h4><p>Generating test coverage (using Vitest or Jest) for pre-architected modules. AI can generate comprehensive unit tests instantly when given clear inputs and expected outputs.</p><h3>Conclusion: AI Amplifies Your Architecture, It Doesn’t Replace It</h3><p>AI will not make a bad architect build good software — it will simply help a bad architect produce spaghetti code ten times faster.</p><p>The software engineers who thrive in this new era are not those who delegate all thinking to LLMs. They are the ones who protect their foundational knowledge , take responsibility for architecture and maintainability , and use AI as a high-speed execution engine.</p><p>Let the AI handle the typing, but make sure you are always the one steering the architecture.</p><h3>About the Author</h3><p><strong>Muhammad Azmi</strong> is a Software Engineer and Frontend Developer from Indonesia with 3+ years of professional experience. His work spans from building complex Electronic Medical Record (EMR) systems for public hospitals to crafting high-performance landing pages for global brands like Infinix and full-scale ERP systems.</p><p>Want to explore my selected production projects, articles, and full-stack architecture? Check out my portfolio at <a href="https://muhammadazmi.web.id/en">muhammadazmi.web.id</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=af9daff402d2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Stop Seeing Ads on Your Phone: The Complete Guide to DNS Ad Blocking (Android & iOS)]]></title>
            <link>https://medium.com/@mhmmdazmye/stop-seeing-ads-on-your-phone-the-complete-guide-to-dns-ad-blocking-android-ios-4f0752246dda?source=rss-05407ce5c0ef------2</link>
            <guid isPermaLink="false">https://medium.com/p/4f0752246dda</guid>
            <category><![CDATA[ios]]></category>
            <category><![CDATA[website]]></category>
            <category><![CDATA[android]]></category>
            <dc:creator><![CDATA[Muhammad Azmi]]></dc:creator>
            <pubDate>Fri, 24 Jul 2026 16:06:39 GMT</pubDate>
            <atom:updated>2026-07-24T16:06:39.631Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*FoYxgLJxNzRUeT6ga7l1bg.png" /><figcaption>AI Generated Image</figcaption></figure><p>A simple, system-wide solution to block ads without draining your battery or installing sketchy apps.</p><p>Browsing the web on a mobile device can sometimes feel like navigating a minefield. Ads and trackers bloat web pages, push down content, cause frustrating layout shifts, and quietly drain your mobile data in the background.</p><p>While you could install a dedicated ad-blocking browser or a background VPN app, those solutions often consume excess memory and drain your battery faster. The most elegant and efficient way to block ads system-wide — across both your mobile browsers and your installed apps — is by changing your <strong>DNS (Domain Name System)</strong> routing.</p><p>Here is the exact, step-by-step guide on how to seamlessly block ads on both Android and iOS devices using AdGuard’s free public DNS.</p><h3>What is a DNS Ad Blocker?</h3><p>Think of DNS as the “phonebook” of the internet. When you type google.com (or when an app fetches data), your DNS server looks up the IP address for that domain so your phone can load it.</p><p>When you use an ad-blocking DNS, this “phonebook” works normally for regular websites. But when a website or app attempts to resolve a known ad-serving domain (like ads.doubleclick.net), the DNS server simply drops the request. The ad never loads, no screen space is taken up, and your data is saved.</p><h3>How to Set Up Ad-Blocking DNS on Android</h3><p>Android has natively supported encrypted DNS (DNS-over-TLS) for years, making this setup incredibly simple. You can configure it directly from your system settings without downloading any third-party software.</p><p><strong>Step-by-Step Instructions:</strong></p><ol><li>Open your phone’s <strong>Settings</strong>.</li><li>Navigate to <strong>Network &amp; Internet</strong> (on some devices, like Samsung, this is found under <strong>Connections</strong> → <strong>More connection settings</strong>).</li><li>Tap on <strong>Private DNS</strong>.</li><li>Select the option for <strong>Private DNS provider hostname</strong> (sometimes labeled “Designated private DNS”).</li><li>In the text field, type the official AdGuard host address: dns.adguard-dns.com <em>(Note: </em><em>dns.adguard.com also works).</em></li><li>Tap <strong>Save</strong>.</li></ol><p><strong>The Result:</strong> You are now connected. You will immediately notice that in-app banner ads, pop-ups, and video ads on websites will fail to load, leaving a much cleaner and faster experience.</p><h3>How to Set Up Ad-Blocking DNS on iOS (iPhone &amp; iPad)</h3><p>Unlike Android, iOS doesn’t have a simple text box in the main settings to type a custom hostname for all your connections. However, Apple natively supports encrypted DNS (DNS-over-HTTPS/TLS) via Configuration Profiles or manual Wi-Fi adjustments.</p><p>Here are the two ways to set it up:</p><h4>Method 1: The Profile Method (Recommended for System-Wide Blocking)</h4><p>If you want to block ads on both your Cellular Data (4G/5G) and Wi-Fi, downloading a Configuration Profile is the best route.</p><ol><li>Open <strong>Safari</strong> on your iPhone and go to AdGuard’s public configuration page: adguard-dns.io/en/public-dns.html.</li><li>Scroll down to the iOS section and tap to <strong>Download the configuration profile</strong> for the “Default Server”.</li><li>A popup will ask if you want to allow the download. Tap <strong>Allow</strong>.</li><li>Now, open your iPhone’s <strong>Settings</strong> app. You will see a new menu item near the top that says <strong>Profile Downloaded</strong>. Tap it.</li><li>Tap <strong>Install</strong> in the top right corner and enter your iPhone passcode.</li><li>Follow the final on-screen prompts to finish the installation.</li></ol><p><em>Pro-Tip: To manage this later or temporarily turn it off, simply go to </em><strong><em>Settings &gt; General &gt; VPN, DNS, &amp; Device Management &gt; DNS</em></strong><em>.</em></p><h4>Method 2: The Manual Wi-Fi Method</h4><p>If you prefer not to install a profile, you can manually change your DNS. However, <strong>this method only applies to the specific Wi-Fi network you configure it on</strong>, meaning ads will return the moment you switch to cellular data.</p><ol><li>Open <strong>Settings</strong> &gt; <strong>Wi-Fi</strong>.</li><li>Tap the blue <strong>“i” (info) icon</strong> next to your currently connected Wi-Fi network.</li><li>Scroll down to the bottom and tap <strong>Configure DNS</strong>.</li><li>Change the setting from Automatic to <strong>Manual</strong>.</li><li>Delete any existing DNS servers by tapping the red minus button.</li><li>Tap <strong>Add Server</strong> and enter AdGuard’s public IPv4 addresses: 94.140.14.14 and 94.140.15.15</li><li>Tap <strong>Save</strong> in the top right corner.</li></ol><h3>Conclusion: Enjoy a Cleaner, Faster Mobile Experience</h3><p>By utilizing custom DNS routing on your personal devices, you ensure that your phone only loads the content you actually want to see — free from the heavy, bloated scripts of third-party ad networks.</p><p>Try it out for just a few days. You will be shocked at how much faster your device feels, how much cleaner your apps look, and how much mobile data you save in the process.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4f0752246dda" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Stop Coding for Crawlers: Outdated SEO Tactics Web Developers Need to Drop in 2026]]></title>
            <link>https://medium.com/@mhmmdazmye/stop-coding-for-crawlers-outdated-seo-tactics-web-developers-need-to-drop-in-2026-43b2c542dbd4?source=rss-05407ce5c0ef------2</link>
            <guid isPermaLink="false">https://medium.com/p/43b2c542dbd4</guid>
            <category><![CDATA[amp]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[seo-tips]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Muhammad Azmi]]></dc:creator>
            <pubDate>Wed, 22 Jul 2026 18:19:21 GMT</pubDate>
            <atom:updated>2026-07-22T18:19:21.319Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*KA3hJuZd22FJxyprUyXc9w.png" /><figcaption>AI Generated Image</figcaption></figure><p><strong>Why focusing on user experience and modern architecture beats chasing dead search engine algorithms.</strong></p><p>As a frontend developer, I’ve spent the last few years building everything from complex Electronic Medical Record (EMR) systems for public hospitals to high-performance, animation-heavy landing pages for global brands like Infinix. If there is one thing I’ve learned about bridging the gap between clean code and business value, it’s this: <strong>Performance is paramount.</strong></p><p>But there is a lingering ghost in the web development industry.</p><p>Many developers and “old-school” SEO specialists are still relying on a playbook written a decade ago. They treat Search Engine Optimization as a dark art of tricking crawlers, rather than what it actually is in 2026: building fast, accessible, and technically sound web architectures.</p><p>If you want your web applications to rank today, you have to stop coding for crawlers and start coding for users. Here are two massive, outdated SEO tactics that developers need to completely drop right now.</p><h3>1. The Meta Keywords Tag (The Zombie of SEO)</h3><p>It is baffling how often I still look at the &lt;head&gt; of a newly deployed website and see this:</p><pre>&lt;meta name=&quot;keywords&quot; content=&quot;best web developer, software engineer jakarta, react developer, buy cheap shoes, etc&quot;&gt;</pre><p><strong>The Old Way:</strong> In the early 2000s, search engines weren’t smart enough to read the context of a page, so developers had to explicitly tell them what the page was about using the meta keywords tag.</p><p><strong>Why it Hurts Now:</strong> Google officially announced that it completely ignores the meta keywords tag back in <strong>2009</strong>. That was nearly two decades ago. Google’s Natural Language Processing (NLP) and semantic search algorithms now read the actual content of your page to understand user intent.</p><blockquote>”Our web search (the well-known search at Google.com that hundreds of millions of people use each day) disregards keyword metatags completely. They simply don’t have any effect in our search ranking at present.”— <a href="https://developers.google.com/search/blog/2009/09/google-does-not-use-keywords-meta-tag">Google Search Central</a></blockquote><p>But it gets worse. Not only does Google ignore it, but Bing has officially stated that they use the meta keywords tag as a <strong>spam signal</strong>. If you stuff dozens of keywords into this tag, search engines assume you are trying to game the system. You aren’t helping your SEO; you are actively triggering spam filters and wasting precious bytes in your initial HTML payload.</p><blockquote>“The meta keyword tag is dead in terms of SEO value. Sure, it might have value for contextual ad systems or act as a signal to bots plying the web looking for topics to target, but as far as search goes, that tag flat out tanked years ago as a booster. Today, it’s pretty much just a spam signal.” — <a href="https://blogs.bing.com/webmaster/October-2014/Blame-The-Meta-Keyword-Tag"><em>Duane Forrester, Bing</em></a></blockquote><p><strong>The Fix:</strong> Delete it. Completely. Focus instead on writing highly semantic HTML5, using proper &lt;h1&gt; to &lt;h6&gt; hierarchies, and crafting compelling &lt;title&gt; and &lt;meta name=&quot;description&quot;&gt; tags that actually make a user want to click your link in the search results.</p><h3>2. AMP (Accelerated Mobile Pages)</h3><p><strong>The Old Way:</strong> A few years ago, Google pushed AMP hard. The promise was simple: use our restrictive, stripped-down HTML framework, and we will serve your pages lightning-fast from our cache and prioritize you in the “Top Stories” mobile carousel. Developers scrambled to maintain two separate codebases — one for their main site and one for AMP.</p><p><strong>Why it Hurts Now:</strong> Google has quietly dismantled the AMP requirement. It is no longer required to appear in Top Stories, and the visual badge has been removed from search results.</p><blockquote>“As part of this update, we’ll also incorporate the page experience metrics into our ranking criteria for the Top Stories feature in Search on mobile, and remove the AMP requirement from Top Stories eligibility. Google no longer shows the AMP badge icon to indicate AMP content.” — <a href="https://developers.google.com/search/blog/2020/05/evaluating-page-experience#:~:text=As%20part%20of%20this%20update%2C%20we%27ll%20also%20incorporate%20the%20page%20experience%20metrics%20into%20our%20ranking%20criteria%20for%20the%20Top%20Stories%20feature%20in%20Search%20on%20mobile%2C%20and%20remove%20the%20AMP%20requirement%20from%20Top%20Stories%20eligibility."><em>Google Search Central</em></a></blockquote><p>Maintaining an AMP version of a site today is a massive drain on developer resources. It forces you to abandon modern JavaScript functionality, restricts your CSS, and creates a fractured codebase. More importantly, it is entirely unnecessary.</p><p>In 2026, Core Web Vitals (Largest Contentful Paint, First Input Delay, Cumulative Layout Shift) are the metrics that actually matter. You do not need Google’s proprietary framework to achieve perfect scores.</p><p><strong>The Fix:</strong> Rely on modern frontend architecture. Frameworks like Next.js and Nuxt.js (which I actively use for high-traffic landing pages) provide out-of-the-box solutions like Static Site Generation (SSG) and Server-Side Rendering (SSR). By leveraging standard modern practices — image optimization, code splitting, edge caching, and minimizing main-thread blocking JavaScript — you can build standard web pages that load instantly and vastly outperform legacy AMP pages, without sacrificing the user experience or complex UI animations.</p><h4>Conclusion: Build Products That Sell Themselves</h4><p>Great SEO isn’t about tricking algorithms with hidden tags or proprietary mobile frameworks. It’s about building accessible, fast, and cleanly structured web applications that clearly communicate their value to both humans and machines.</p><p>When you build an architecture that loads instantly, doesn’t shift around while reading, and serves clean HTML on the first request, the search engine algorithms will naturally reward you.</p><p>Stop coding for crawlers. Build for humans. The market will find you.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=43b2c542dbd4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI is a Co-Pilot, Not an Autopilot: How to Keep Your Software Engineering Soul]]></title>
            <link>https://medium.com/@mhmmdazmye/ai-is-a-co-pilot-not-an-autopilot-how-to-keep-your-software-engineering-soul-3fe05cc75e54?source=rss-05407ce5c0ef------2</link>
            <guid isPermaLink="false">https://medium.com/p/3fe05cc75e54</guid>
            <category><![CDATA[future-of-work]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Muhammad Azmi]]></dc:creator>
            <pubDate>Mon, 20 Jul 2026 21:04:45 GMT</pubDate>
            <atom:updated>2026-07-20T21:04:45.918Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*c_e3RmhpIARRGPNpQnoMoA.png" /></figure><p>It feels incredibly good to press Tab and watch GitHub Copilot spit out a perfectly formatted 50-line React component in a fraction of a second. It feels like magic. It feels like the future.</p><p>But if we aren’t careful, it is also a trap.</p><p>In my previous article on <a href="https://medium.com/@mhmmdazmye/the-ai-trap-why-the-tech-industry-will-soon-beg-for-junior-developers-fe7f50386867?sharedUserId=mhmmdazmye">how AI is breaking the developer pipeline</a>, I talked about the macro-level industry problem: companies aren’t hiring juniors, which cuts off the lifecycle of future senior engineers. Today, I want to talk about the micro-level problem — the one happening right inside your IDE.</p><p>We are facing an epidemic of “cognitive atrophy” in software engineering. When you outsource the struggle of debugging, state management, and algorithmic thinking to an LLM, you are outsourcing your own growth.</p><p>If you want to survive and thrive in this new era, you need to establish strict boundaries with your AI tools. Here are my four rules of engagement for using AI without losing your software engineering soul.<br>The Cognitive Atrophy Trap</p><p>Building software is hard. It requires a specific type of mental elasticity that is only forged through frustration. The hours spent wrestling with a stubborn API response or a weird React re-render cycle are exactly how developers build “muscle memory.”</p><p>When you bypass that struggle entirely, your brain gets lazy. You stop thinking about <em>how</em> systems connect and start acting as a copy-paste middleman between ChatGPT and VS Code. Eventually, when a complex, production-breaking bug occurs that AI cannot solve, you will find yourself completely paralyzed.</p><p>To prevent this, you have to treat AI as what it is: a co-pilot, not an autopilot.</p><h3>Rule 1: Never merge code you cannot explain line-by-line.</h3><p>It’s tempting to let AI write a complex Regex pattern or an intricate SQL query, shrug, and say, “Well, it works.”</p><p>Never do this. If an AI writes a block of code, you must be able to stand in front of a whiteboard and explain its logic to your team. If it introduces a subtle security vulnerability or a massive memory leak in production, “the AI wrote it” will not save your job. You are the engineer; you hold the liability.</p><h3>Rule 2: Outsource the typing, not the architecture.</h3><p>AI is phenomenal at tedious tasks: mapping over arrays, generating boilerplate layouts, or writing repetitive CSS/Tailwind classes. Let it handle the grunt work.</p><p>But reserve your human brainpower for the high-level architecture: system design, component hierarchy, security, and data flow.</p><p>For example, when I was building a complex <strong>Electronic Medical Record (EMR) system for a public hospital</strong>, I used AI to speed up the mundane UI boilerplate. But I relied entirely on my own engineering fundamentals to structure the React state management and handle the critical patient-data architecture. You cannot afford to let an LLM hallucinate logic when real-world stakes — like healthcare data — are involved.</p><h3>Rule 3: Use AI as a tutor, not just a generator.</h3><p>Change the way you prompt. Instead of asking, <em>“Write a function to sort this data by date,”</em> try asking, <em>“Explain the most efficient algorithm to sort this specific data structure in JavaScript, and give me a small hint so I can write it myself.”</em></p><p>This simple shift changes AI from a crutch into a personalized senior mentor. It forces you to engage with the logic, ensuring that your core coding skills stay sharp while still getting the answers you need quickly.</p><h3>Rule 4: Break the testing echo chamber.</h3><p>If AI writes your component <em>and</em> you ask it to write the unit tests for that same component, you are creating a dangerous feedback loop. The AI will often simply validate its own flawed logic.</p><p>If the AI misunderstood the edge cases during implementation, it will ignore those same edge cases in the tests. Always write your own test cases, or at the very least, heavily audit and expand upon the ones the AI generates to ensure they test actual business requirements, not just the code’s syntax.</p><h3>The Bottom Line</h3><p>AI is not going to replace software engineers, but software engineers who use AI poorly will be replaced by those who use it with discipline.</p><p>Protect your struggle. Protect your foundational knowledge. Let the AI type, but make sure you are always the one driving.</p><p><em>Want to see how I build resilient, modern tech stacks in the real world? </em><a href="https://muhammadazmi.web.id/en"><em>Check out my portfolio and selected projects here</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3fe05cc75e54" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The AI Trap: Why the Tech Industry Will Soon Beg for Junior Developers]]></title>
            <link>https://medium.com/@mhmmdazmye/the-ai-trap-why-the-tech-industry-will-soon-beg-for-junior-developers-fe7f50386867?source=rss-05407ce5c0ef------2</link>
            <guid isPermaLink="false">https://medium.com/p/fe7f50386867</guid>
            <category><![CDATA[future-of-work]]></category>
            <category><![CDATA[software-engineering]]></category>
            <category><![CDATA[careers]]></category>
            <dc:creator><![CDATA[Muhammad Azmi]]></dc:creator>
            <pubDate>Sun, 19 Jul 2026 21:18:45 GMT</pubDate>
            <atom:updated>2026-07-20T10:58:44.734Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="The AI Trap: Why the Tech Industry Will Soon Beg for Junior Developers" src="https://cdn-images-1.medium.com/max/1024/1*q7QIQyNenXXLMaHgZPTbmg.png" /></figure><p>AI is doing the heavy lifting for entry-level tasks right now. But breaking the software engineering lifecycle comes with a massive future cost.</p><p>If you look at the current tech landscape, the narrative is everywhere: AI is writing our code. With the rapid evolution of large language models and AI coding assistants, the day-to-day reality of writing software has fundamentally changed. The immediate casualty in this new era? The Junior Developer.</p><p>Currently, many companies have quietly paused or significantly reduced their hiring for entry-level roles. Why pay a junior engineer to write boilerplate code, fix minor UI bugs, or draft basic components when an AI can do it in seconds at a fraction of the cost? It seems like a smart business move in the short term. But zoom out, and you will see a massive, ticking time bomb underneath the industry: we are actively cutting the software engineering lifecycle.</p><h3>The Broken Lifecycle</h3><p>To understand the crisis, we have to look at how great software engineers are made. Senior developers do not spawn out of thin air. They are the product of years of making mistakes, taking down production environments, debugging legacy spaghetti code, and learning how to communicate with stakeholders. It takes time and hands-on experience to learn how to build complex, real-world applications — whether that is an <a href="https://muhammadazmi.web.id/en">electronic medical record system for a public hospital</a> or a comprehensive <a href="https://muhammadazmi.web.id/en">ERP system</a>.</p><p>When a junior developer starts, they aren’t just there to push code; they are there to learn the context of software. They learn that a successful product isn’t just about making it work, but ensuring that each project is built to hold up in the details, from typography to performance.</p><p>By offloading junior-level tasks to AI, companies are effectively breaking the pipeline. If nobody is hired to do the junior work today, who will be the mid-level engineers in three years? Who will be the senior architects in ten?</p><h3>The Illusion of “AI as a Senior Developer”</h3><p>There is a dangerous misconception that AI will eventually just replace senior developers, too. But writing syntax is only about 20% of a senior engineer’s job.</p><p>The real value of an experienced developer lies in:</p><ul><li><strong>System Architecture:</strong> Understanding how different microservices interact without bottlenecking.</li><li><strong>Business Translation:</strong> Turning vague client requirements into logical, scalable technical realities.</li><li><strong>Edge-Case Anticipation:</strong> Knowing what happens when a user does something entirely unpredictable that an AI didn’t foresee in its training data.</li></ul><p>AI is an incredibly powerful autocomplete, but it is not a software engineer. It doesn’t sit in cross-functional meetings, it doesn’t push back against bad product requirements, and it doesn’t take accountability when a deployment fails. Companies will always need senior human engineers to oversee, architect, and take responsibility for the final product.</p><p>But without a junior pipeline, the pool of available senior talent will soon dry up, leading to an unprecedented hiring crisis.</p><h3>The Inevitable Reversal: Why Companies Will Need Juniors Again</h3><p>Within the next few years, the tech industry is going to hit a wall. As the current crop of senior engineers retires, moves into management, or leaves the industry, companies will face a severe talent drought. The cost of hiring the few remaining experienced engineers will skyrocket.</p><p>Out of sheer necessity, companies will be forced to start hiring junior developers again. However, the nature of the “junior” role will look entirely different.</p><h4>The “New” Junior Developer</h4><p>Instead of spending their first year writing boilerplate, the junior developers of the AI era will become something more akin to AI Operators or Junior Reviewers. Their daily responsibilities will shift:</p><ol><li><strong>Prompt Engineering &amp; AI Management:</strong> Generating the baseline code using AI tools, and then heavily auditing it for security, performance, and business logic.</li><li><strong>Focusing on the “Why” Over the “How”:</strong> Because AI handles the syntax (the “how”), juniors will be forced to learn system design and architecture (the “why”) much earlier in their careers.</li><li><strong>Quality Assurance &amp; Details:</strong> Ensuring the AI-generated output actually meets human standards — ironing out the UX quirks, perfecting the typography, and optimizing load times to ensure the product truly holds up.</li></ol><h3>A Call to the Tech Industry</h3><p>We cannot afford to sever the roots of our own industry. If companies want a sustainable technical future, they need to view junior developers not as an immediate ROI, but as an investment in their future architecture.</p><p>The cycle of software engineering has been disrupted, but it isn’t beyond repair. The companies that realize this now — and continue to hire, mentor, and integrate juniors alongside AI tools — will be the ones holding all the senior talent when the rest of the industry is left scrambling.</p><h3>About the Author</h3><p><strong>Muhammad Azmi</strong> is a Software Engineer and Frontend Developer from Indonesia with over 3 years of professional experience. His work encompasses building complex electronic medical record systems for public hospitals, crafting dynamic landing pages, and developing ERP systems at a software house. Every project he touches is meticulously built to hold up in the details, prioritizing everything from typography to performance.</p><p><strong>Explore his selected projects and professional journey:</strong> <a href="https://muhammadazmi.web.id/en"><strong>https://muhammadazmi.web.id/en</strong></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fe7f50386867" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[SEO for Web Developers in 2026: This is How You Sell Your Product]]></title>
            <link>https://medium.com/@mhmmdazmye/seo-for-web-developers-in-2026-this-is-how-you-sell-your-product-78a7bc67074b?source=rss-05407ce5c0ef------2</link>
            <guid isPermaLink="false">https://medium.com/p/78a7bc67074b</guid>
            <category><![CDATA[technology]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[marketing]]></category>
            <dc:creator><![CDATA[Muhammad Azmi]]></dc:creator>
            <pubDate>Sun, 19 Jul 2026 18:17:20 GMT</pubDate>
            <atom:updated>2026-07-22T20:57:12.568Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ThImZIAXoh4OAddrrK11aw.png" /><figcaption>AI Generated Image</figcaption></figure><p><strong>In 2026, the battle for web developers isn’t just about building the perfect product — it’s about putting it on the market and making sure it actually gets found.</strong></p><p>We’ve all been there. You spend weeks architecting the perfect full-stack app. The database schema is flawless, the API routes are snappy, and the frontend transitions are buttery smooth. You deploy it, share the link on a few Discord servers, and wait for the users to roll in.</p><p>And then… crickets.</p><p>The harsh reality of building on the web today is that engineering brilliance does not equal traffic. If you want to sell your SaaS, land freelance clients, or get users for your side project, Search Engine Optimization (SEO) can no longer be an afterthought. It has to be part of your architecture.</p><p>Here is how modern developers need to approach technical SEO to actually sell their products.</p><h3>1. Code for AI, Not Just Browsers (Semantic HTML)</h3><p>Search engines and LLM crawlers do not care how beautiful your UI is. They parse the DOM to understand the hierarchy and relationships of your data. If you are still throwing &lt;div&gt; tags at every layout problem, you are blinding the bots that dictate your traffic.</p><p>When you use semantic HTML, you are essentially providing a clean API for AI crawlers to understand your product’s value proposition.</p><ul><li>Use &lt;main&gt; for the core feature set.</li><li>Use &lt;article&gt; for your blog or use-case studies.</li><li>Use &lt;nav&gt; for routing links so crawlers understand site architecture.</li></ul><p>If an AI overview cannot confidently pull the answer to “What does this software do?” because the text is buried in nested, meaningless tags, it will recommend your competitor’s product instead.</p><h3>2. Talk Directly to the Algorithm (JSON-LD)</h3><p>If you want rich snippets — like star ratings, product prices, or feature summaries — showing up in search results, you must use Schema Markup via JSON-LD.</p><p>This is where you explicitly hand search engines a JSON object detailing exactly what you are selling. In a modern React or Vite application, you can build a reusable component to inject this context directly into the DOM.</p><pre>// A simple reusable JSON-LD component for React<br>export function JsonLd({ data }) {<br>  return (<br>    &lt;script<br>      type=&quot;application/ld+json&quot;<br>      dangerouslySetInnerHTML={{ __html: JSON.stringify(data) }}<br>    /&gt;<br>  );<br>}</pre><p>By passing a standard SoftwareApplication or Product schema object into this component, you tell Google exactly what your app is, how much it costs, and who it’s for.</p><h3>3. Speed is a Sales Metric (Core Web Vitals)</h3><p>Google explicitly ranks sites based on user experience metrics (Core Web Vitals), but more importantly, slow sites kill conversions.</p><ul><li><strong>Largest Contentful Paint (LCP):</strong> If your hero section doesn’t load instantly, the user bounces.</li><li><strong>Cumulative Layout Shift (CLS):</strong> If your “Buy Now” button jumps around while loading, you lose the sale.</li></ul><p>Heavy JavaScript bundles ruin these metrics. Furthermore, backend latency matters just as much. If your server takes 800ms to respond, your LCP fails. This is why deploying lightweight backends (like Hono.js) on edge networks (like Cloudflare Workers) alongside edge databases (like Turso) isn’t just a fun technical exercise — it’s a direct competitive advantage for your product’s visibility.</p><h3>4. Stop Hiding Your Content Behind Client-Side Rendering</h3><p>Client-Side Rendering (CSR) via standard React or Vue is notorious for causing SEO headaches. While Googlebot <em>can</em> execute JavaScript, it often queues JS-heavy pages to be rendered days later. If you are launching a product, you can’t afford that delay.</p><p>To ensure your landing pages and product features are indexed instantly:</p><ul><li><strong>Static Site Generation (SSG):</strong> Best for marketing pages where the core pitch doesn’t change by the second.</li><li><strong>Server-Side Rendering (SSR):</strong> Best for dynamic storefronts or directories that need to be instantly readable by crawlers upon request.</li></ul><p>If you are shipping a standard Single Page Application (SPA), ensure you are using a prerendering tool or a framework adapter so bots immediately see your value proposition.</p><h3>Conclusion: Ship Products That Sell Themselves</h3><p>Great SEO isn’t about tricking algorithms with keyword stuffing. It’s about building accessible, fast, and cleanly structured web applications that clearly communicate their value to both humans and machines.</p><p>I recently put these performance and architectural principles to the test while building my custom landing page. If you want to see how this translates into a live, production-grade environment, or want to connect regarding your next project, check out my portfolio at <a href="https://muhammadazmi.web.id/en">muhammadazmi.web.id</a>.</p><p>Build it fast, structure it right, and the market will find you.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=78a7bc67074b" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>