<?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 Nobert Momoh on Medium]]></title>
        <description><![CDATA[Stories by Nobert Momoh on Medium]]></description>
        <link>https://medium.com/@nmxmxh?source=rss-1ff1d334efdb------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*fgn5KiyUIDmLy0rN-tOM9A.jpeg</url>
            <title>Stories by Nobert Momoh on Medium</title>
            <link>https://medium.com/@nmxmxh?source=rss-1ff1d334efdb------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Wed, 18 Nov 2026 06:41:26 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@nmxmxh/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[ Systematic Debugging Prompt Structure]]></title>
            <link>https://nmxmxh.medium.com/systematic-debugging-prompt-structure-9e4ed3354c76?source=rss-1ff1d334efdb------2</link>
            <guid isPermaLink="false">https://medium.com/p/9e4ed3354c76</guid>
            <category><![CDATA[writing-prompts]]></category>
            <category><![CDATA[systems-thinking]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[prompt-engineering]]></category>
            <category><![CDATA[debugging]]></category>
            <dc:creator><![CDATA[Nobert Momoh]]></dc:creator>
            <pubDate>Thu, 25 Sep 2025 21:24:52 GMT</pubDate>
            <atom:updated>2025-09-25T21:24:52.864Z</atom:updated>
            <content:encoded><![CDATA[<p>Phase 1: Problem Understanding &amp; Context<br>```<br>Before making any changes, I need to understand:</p><p>1. What exactly is broken?<br> — Describe the specific symptoms<br> — What should be happening vs what is happening?<br> — Provide logs/evidence of the issue</p><p>2. What was working before?<br> — When did this last work correctly?<br> — What changed recently?<br> — Are there similar features that work?</p><p>3. Architecture constraints:<br> — Don’t create new files unless absolutely necessary<br> — Focus on streamlining existing architecture<br> — Maintain consistency across the system<br> — Find root causes, not symptoms<br>```</p><p>Phase 2: System Investigation<br>```<br>Now investigate systematically:</p><p>1. Map the entire data/event flow:<br> — Trace from source to destination<br> — Identify all components involved<br> — Don’t assume where the problem is</p><p>2. Compare with working examples:<br> — How do similar services handle this?<br> — What patterns work elsewhere in the codebase?<br> — What’s different about the broken case?</p><p>3. Examine the architecture:<br> — What are the canonical patterns?<br> — How should this work according to the system design?<br> — What are the established conventions?<br>```</p><p>Phase 3: Root Cause Analysis<br>```<br>Before proposing any fix:</p><p>1. Identify the real root cause:<br> — Where in the flow does the data get lost/corrupted?<br> — What component is behaving differently than expected?<br> — Why is it behaving that way?</p><p>2. Validate the hypothesis:<br> — Can you trace the exact point of failure?<br> — Does this explain all the symptoms?<br> — Is this consistent with the working examples?</p><p>3. Consider system-wide impact:<br> — How would this fix affect other services?<br> — Does this maintain architectural consistency?<br> — Are there unintended consequences?<br>```</p><p>Phase 4: Solution Design<br>```<br>Design the fix:</p><p>1. Architectural approach:<br> — How can we fix this using existing patterns?<br> — What’s the minimal change that solves the root cause?<br> — How does this improve the overall system?</p><p>2. Implementation strategy:<br> — Modify existing files, don’t create new ones<br> — Use established patterns and conventions<br> — Ensure the fix is generic, not an edge case</p><p>3. Validation plan:<br> — How will we verify this fixes the issue?<br> — How will we ensure it doesn’t break other things?<br> — What tests or checks should we run?<br>```</p><p>Phase 5: Implementation &amp; Verification<br>```<br>Execute the fix:</p><p>1. Make the minimal necessary changes:<br> — Focus on the root cause<br> — Use existing architectural patterns<br> — Don’t over-engineer</p><p>2. Verify the fix:<br> — Test the specific issue<br> — Ensure other services still work<br> — Check that the architecture is more consistent</p><p>3. Document the learning:<br> — What was the real root cause?<br> — How does this improve the system?<br> — What patterns should we follow going forward?<br>```</p><p>🚫 Anti-Patterns to Avoid</p><p>```<br>❌ DON’T:<br>- Jump to solutions without understanding the system<br>- Create new files when existing ones can be modified<br>- Fix symptoms instead of root causes<br>- Make changes specific to one service without considering others<br>- Assume where the problem is without investigating<br>- Rush to implement without understanding the architecture</p><p>✅ DO:<br>- Understand the entire system first<br>- Trace the complete data flow<br>- Compare with working examples<br>- Find the root cause, not symptoms<br>- Use existing architectural patterns<br>- Make changes that improve the overall system<br>```</p><p>📋 Quick Reference Checklist</p><p>```<br>Before making any changes, ask:<br>□ Have I traced the entire data/event flow?<br>□ Have I compared with working examples?<br>□ Do I understand the architectural patterns?<br>□ Have I identified the root cause?<br>□ Will this fix improve the overall system?<br>□ Am I using existing patterns, not creating new ones?<br>□ Does this maintain consistency across services?<br>```</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9e4ed3354c76" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Supercomputers Everywhere: WASM, Ambition, and the Global South]]></title>
            <link>https://nmxmxh.medium.com/supercomputers-everywhere-wasm-ambition-and-the-global-south-3f0ce655d59b?source=rss-1ff1d334efdb------2</link>
            <guid isPermaLink="false">https://medium.com/p/3f0ce655d59b</guid>
            <category><![CDATA[webassembly]]></category>
            <category><![CDATA[distributed-systems]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[golang]]></category>
            <dc:creator><![CDATA[Nobert Momoh]]></dc:creator>
            <pubDate>Wed, 20 Aug 2025 14:44:25 GMT</pubDate>
            <atom:updated>2025-08-20T14:44:25.750Z</atom:updated>
            <content:encoded><![CDATA[<h3>Introduction</h3><p>One of my childhood dreams, was to build a supercomputer. I was motivated by the grand awe of building a massive computing machine, to do all sorts of sciency and futuristic stuff.</p><p>Growing up in the global south, the dream of building a supercomputer often felt out of reach. The barriers — location, resources, infrastructure — seemed insurmountable. Yet, for those with the eyes to see, there are supercomputers everywhere. The world is changing, and so are the tools at our disposal.</p><h3>The Philosophy:</h3><p><strong>Ambition vs. Location</strong></p><p>Ambition is universal, but opportunity is not. The global south is rich in talent, but often poor in hardware. Traditional supercomputers — massive, centralized, chip-heavy — are rare. But what if we could sidestep these limits? What if the network itself became our supercomputer?</p><h3>WASM: The Great Equalizer</h3><p>WebAssembly (WASM) is more than a technology — it’s a philosophy. It allows code to run anywhere: browsers, servers, edge devices. In this <a href="https://github.com/nmxmxh/master-ovasabi">codebase</a>, WASM modules (see <a href="https://github.com/nmxmxh/master-ovasabi/blob/main/wasm/main.go">main.go</a>, <a href="https://github.com/nmxmxh/master-ovasabi/blob/main/frontend/public/main.wasm">main.wasm</a>) act as bridges, letting Go code execute efficiently across platforms.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pkw0Z_OMxzZG9Es157y_bA.png" /></figure><h3>Beyond the Chip: Distributed AI</h3><p>AI is often bottlenecked by hardware — GPUs, TPUs, expensive chips. But WASM lets us rethink this. Imagine a network of devices — phones, laptops, edge nodes — all running WASM modules, all contributing compute. The supercomputer is no longer a place; it’s a network.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HnbtG5EflW2XIcL1TuimdA.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*pQ25syuM07S8h5OkNnlf8A.png" /></figure><h3>The Vision:</h3><p><strong>Supercomputers Everywhere</strong></p><p>With WASM, every device is a node. Every browser is a gateway. The global south, with its millions of connected devices, can become a supercomputer — if we build the right bridges.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ECicgyNTteV6oH9Zrkwzpw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*rYKd0YvDota2kzkU33IOMw.png" /></figure><h3>Conclusion</h3><p>The limits of chip-based AI are real, but not absolute. WASM lets us leverage the network, not just the node. For those with the eyes to see, there are supercomputers everywhere. The future is distributed, and it’s already here.</p><h3>Call to Action</h3><p>If you’re building in the global south, don’t wait for the hardware. Build the bridges. Leverage WASM. The supercomputer is all around you.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=3f0ce655d59b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Turning Setback into Advantage: Why Losing European Football May Benefit Manchester United in the…]]></title>
            <link>https://nmxmxh.medium.com/turning-setback-into-advantage-why-losing-european-football-may-benefit-manchester-united-in-the-68e803ca8ef0?source=rss-1ff1d334efdb------2</link>
            <guid isPermaLink="false">https://medium.com/p/68e803ca8ef0</guid>
            <category><![CDATA[football-manager]]></category>
            <category><![CDATA[premier-league]]></category>
            <category><![CDATA[data-analysis]]></category>
            <category><![CDATA[football]]></category>
            <category><![CDATA[manchester-united]]></category>
            <dc:creator><![CDATA[Nobert Momoh]]></dc:creator>
            <pubDate>Tue, 17 Jun 2025 15:22:42 GMT</pubDate>
            <atom:updated>2025-06-17T15:22:42.915Z</atom:updated>
            <content:encoded><![CDATA[<h3>Turning Setback into Advantage: Why Losing European Football May Benefit Manchester United in the Long Run</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*Bz0fXYJgDB7FgKtH" /><figcaption>Photo by <a href="https://unsplash.com/@dparker07?utm_source=medium&amp;utm_medium=referral">Dan Parker</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><h4>The Initial Disappointment</h4><p>I recall reeling back from my screen disheartened. The club I loved had just lost by a single-goal margin in the Europa League final. It was a dismay performance — not only losing our only hope of attaining silverware in what had been a disappointing season — but also losing our chance of securing a place in next season’s Champion’s League.</p><p>I was heartbroken. I would lose the chance to watch my club perform at the biggest stages, under the brightest lights and, in my own selfishness, the chance to see them play more often. But this also brought up a conversation in my head: at what cost?</p><p>The performance was lacklustre; our players couldn’t generate the required intensity and attitude for a game of that magnitude and essentiality. This had become a constant of our season — players mentally switching off, not having the legs to carry them through the game. But rather than dwell in denial, my curiosity was piqued. I wondered why this had become a constant of Manchester United’s gameplay over recent years.</p><p><strong>NOTE: This is a piece I stumbled upon out of curiosity and not a definitive assertation. I was intrigued by the underlying numbers and thought it’d be fun to create an article like this.</strong></p><p>I will be doing a casual analysis on this issue, focusing on the traditional big six (6) teams in the English Premier League, in relation to minutes, games played, squad depth, and more over the previous 5 seasons — and how this might have affected Manchester United’s ability to cope with the rigorous physical and mental requirements of the modern game, especially in relation to the demands placed on a club of Manchester United’s stature — the biggest in the world.</p><h4>1️⃣ Fixtures Played (Workload)</h4><p>Firstly, let’s take a look at the total matches played over the last 5 seasons for the Big 6 clubs.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*OCNkGlIsZQLrZ2r5Qpy0NQ.png" /><figcaption>24|25 23|24 22|23 21|22 20|21 TOTAL Man Utd 60 52 52 52 52 268 Chelsea 57 51 50 63 59 281 Man City 57 59 61 58 61 296 Liverpool 56 58 52 63 53 292 Tottenham 60 41 50 54 58 263 Arsenal 58 52 49 45 58 262</figcaption></figure><p><strong>SOURCED FROM:</strong> [Transfermarkt]</p><p>This effectively puts Manchester United in the mid range in relation to total match load — not a bad workload, right?</p><h4>2️⃣ Squad Dependence (Player Minutes)</h4><p>Next, let’s take a gander at squad depth. There’s no easy way to properly assess this without combing through all the data, but this is a casual analysis, so I decided to use the average minutes in all competitions for players — noting clubs with players exceeding 2000 minutes and 4000 minutes in a season.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hwZsEmzYv3gGqvf-LyIzRg.png" /><figcaption>24|25 23|24 22|23 21|22 20|21 TOTAL Man Utd 15|4 12|3 15|3 13|1 15|5 70|16 Chelsea 12|0 13|1 10|0 18|2 15|1 68|4 Man City 13|1 13|2 15|3 15|2 17|2 73|10 Liverpool 13|3 17|1 11|2 13|4 10|2 64|12 Tottenham 17|1 11|0 12|1 12|1 14|2 66|5 Arsenal 14|3 11|5 11|1 11|0 15|1 62|10</figcaption></figure><p><strong>SOURCED FROM:</strong> [fbref]</p><p>The statistics immediately stand out. Apart from Manchester City with 73 players having crossed the 2000-minute mark during the season, Manchester United is a close second with 70, and it leads in players exceeding 4000 minutes — 16 in total.</p><h4>3️⃣ Injury Impact (Fitness Levels)</h4><p>This brings us to another point — injuries. There was unfortunately no concrete path to getting information about injury analysis during all the past 5 seasons, with more robust data available only for the last two. Nevertheless, the statistics tell their own story.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EgzNe79K1WWUszXBDP0s1A.png" /><figcaption>24|25 23|24 22|23 21|22 20|21 TOTAL Man Utd 33|1295 45|1620 — — — 78|2915 Chelsea 24|828 43|1745 — — — 67|2573 Man City 33|1134 26|672 — — — 59|1806 Liverpool 22|765 35|1383 — — — 57|2148 Tottenham 41|1553 37|1402 — — — 78|2955 Arsenal 36|1297 23|898 — — — 59|2195</figcaption></figure><p><strong>SOURCED FROM:</strong> [DeepSeek]</p><p>The numbers immediately jump at you. The more volatile clubs — Tottenham and Manchester United — have overwhelming injury numbers, while more stable clubs — Manchester City and Liverpool — show lower figures.</p><h4>4️⃣ Manager, Formation &amp; Club Stability (Variances)</h4><p>This brings up another key consideration — stability. Specifically, variance in league positions, number of managers, formations, and overall mentality. The more variance, the less stability — and a huge amount of pressure is put on the players and backroom to adjust.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*eQZr-uPEyDBn-DJFUVlFnw.png" /><figcaption>M.P V No. of M No. of F Man Utd 6.8 19.04 6 ≈ 8 Chelsea 6.4 10.24 6 ≈ 4 Man City 1.4 0.64 1 ≈ 4 Liverpool 2.4 0.64 2 ≈ 3 Tottenham 7.6 29.44 8 ≈ 5 Arsenal 3.8 6.96 1 ≈ 2</figcaption></figure><p><strong>SOURCED FROM:</strong> [ChatGPT]</p><p>The numbers, once again, jump at you. The clubs with less variance tend to have more stability in relation to their managers, formations, and methodology of play.</p><h4>5️⃣ International Caps (Player Demands)</h4><p>Finally, there was no way to accurately retrieve this information directly (or more accurately, time), so I used ChatGPT to approximate the amount of international caps and appearances for players over the last 5 years.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Bns_1rP4-AfF4hJ2b9hBNg.png" /><figcaption>24|25 23|24 22|23 21|22 20|21 Man Utd ≈ 220 ≈ 210 ≈ 200 ≈ 190 ≈ 180 Chelsea ≈ 200 ≈ 190 ≈ 180 ≈ 170 ≈ 160 Man City ≈ 160 ≈ 150 ≈ 140 ≈ 130 ≈ 120 Liverpool ≈ 150 ≈ 140 ≈ 130 ≈ 120 ≈ 110 Tottenham ≈ 190 ≈ 180 ≈ 170 ≈ 160 ≈ 150 Arsenal ≈ 190 ≈ 180 ≈ 170 ≈ 160 ≈ 150</figcaption></figure><p><strong>NOTE: At this point, I think the model might be leaning heavily on the bias of my inferences, but what can I do?</strong></p><p>The context and impact? Higher cap totals typically translate to greater congestion, less rest, more travel, and physical stress — which collectively undermine performance and consistency. Manchester United’s players frequently find themselves under this pressure, with a high number of internationals adding to their physical and mental workloads.</p><p><strong>NOTE: This also brings up a curious fact, looking into this in relation to the quality and diversity of players in the EPL and how the demand for players by international teams might be affecting rest and recovery periods in relation to other leagues.</strong></p><h4>Summary: Why May Less Be More?</h4><p>This brings me back to the main conclusion. The context you hopefully see here underscores the point I’m trying to make — Manchester United’s struggles are not just a result of poor form or weak character. The team finds itself under enormous physical, mental, and logistical pressure. Players suffer from overdependence, a hectic match calendar, numerous injuries, and constant upheaval in formations, tactics, and leadership. All these factors undermine their ability to perform at their best when it matters most.</p><p>Whilst I will miss the chance to watch my team play more under the brightest European nights, the opportunity to rest, reset, and learn under a tactically aware manager — without the constant pressure and congestion — might be a boon in setting a foundation for future success. Still, losing that final really hurt. It is what it is — but please, Manchester United, reduce the amount of heartbreak you give us.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=68e803ca8ef0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[A Lesson in Code and Philosophy. Falling in love with Go’s Error Handling.]]></title>
            <link>https://nmxmxh.medium.com/a-lesson-in-code-and-philosophy-falling-in-love-with-gos-error-handling-425321623deb?source=rss-1ff1d334efdb------2</link>
            <guid isPermaLink="false">https://medium.com/p/425321623deb</guid>
            <category><![CDATA[error-handling]]></category>
            <category><![CDATA[yin-and-yang]]></category>
            <category><![CDATA[go]]></category>
            <category><![CDATA[computer-science]]></category>
            <category><![CDATA[orchestration]]></category>
            <dc:creator><![CDATA[Nobert Momoh]]></dc:creator>
            <pubDate>Mon, 16 Jun 2025 00:20:27 GMT</pubDate>
            <atom:updated>2025-06-16T00:20:27.576Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SFUu04V0LqzkEAkspsYzaQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QLhP70SnTheV-G02gtCrJw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*lEze2_OweaW-hu0ux5FnxA.png" /></figure><p>While building <a href="https://github.com/nmxmxh/master-ovasabi">INOS</a>, I stumbled across one of Computer Science’s oldest lessons, a great way of addressing it and also falling in love with Go’s error handling.</p><p>Go’s error handling is usually described as a demerit of the language, but I realised, with the right perspective, it’s one of its greatest strengths.</p><p><a href="https://github.com/nmxmxh/master-ovasabi">INOS</a> is metadata driven. Essentially, by shaping the metadata of a specific campaign, the solution will automatically orchestrate the functionality required (think Legos and Minecraft for software).</p><p>This also meant that with successful operations, I had to properly log values, update the metadata and metadata-specific knowledge graph, cache the record, possibly emit an event, and perform other functionality required of a successful operation within the solution. This became somewhat repetitive.</p><p>You’re always advised to DRY in Computer Science. Don’t Repeat Yourself.<br>This made me create Graceful, a package to orchestrate successful operations in the codebase.</p><p>Then I had the thrilling thought to also extend this to the solution’s error handling — essentially a Yin and Yang system — using the package to orchestrate failure as well. This automatically (and unknowingly at the time) elevated the priority and relevance of failures in the system to the same level as success.</p><p>This allowed me to take advantage of Go’s explicit error handling, giving me full visibility and auditability into the errors in my software solution. By treating errors as first-class participants in my application, I had managed to create a system that could fail well and gracefully. This made the code more reliable, easier to debug, and more adaptable in production.</p><p>Take a look at the package here: <a href="https://github.com/nmxmxh/master-ovasabi/tree/main/pkg/graceful">Link</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=425321623deb" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Rediscovering the Internet Operating System: A First-Time Backend Journey]]></title>
            <link>https://nmxmxh.medium.com/rediscovering-the-internet-operating-system-a-first-time-backend-journey-2c3261fa491b?source=rss-1ff1d334efdb------2</link>
            <guid isPermaLink="false">https://medium.com/p/2c3261fa491b</guid>
            <category><![CDATA[google]]></category>
            <category><![CDATA[operating-systems]]></category>
            <category><![CDATA[computer-science]]></category>
            <category><![CDATA[go]]></category>
            <category><![CDATA[distributed-systems]]></category>
            <dc:creator><![CDATA[Nobert Momoh]]></dc:creator>
            <pubDate>Tue, 10 Jun 2025 14:30:44 GMT</pubDate>
            <atom:updated>2025-06-10T14:30:44.575Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*TBfBm0WmEFVIYrxA860MBg.jpeg" /></figure><h3>A Period of Firsts</h3><p>A period of firsts for me.<br> — Building my first backend software solution.<br> — Contributing to open source.<br> — Discovering a 30-year-old Computer Science secret that might’ve stayed buried in the annals of time, and realising I might’ve successfully prototyped its solution — in an era that needs it most.</p><h3>The 30-Year-Old OS Idea</h3><p>Let’s step back a little first.</p><p>Thirty years ago, a different kind of OS was proposed. An <strong>Internet Operating System</strong>, or Internet OS — essentially, an operating system designed to run all of its applications and services through an internet client.</p><p>It would require all its applications to be cross-platform, built on open standards, and not tied to a single user’s computer. Instead, everything — apps, documents, preferences — would live in the internet cloud. It was the perfect vision for what we now call <em>Software as a Service</em>.</p><h3>My Humble Starting Point</h3><p>My initial goal, honestly, wasn’t as grand.<br> I just wanted to build a backend for myself — basing it on everything I’ve learned through projects, jobs, and personal philosophies.</p><p>A colleague recommended <strong>Cursor</strong> to help scaffold the codebase. Before I even realised, I’d overengineered myself into the point of no return.</p><h3>Building Blocks of the System</h3><p>— <strong>Go (Golang)</strong> became the system backbone — think of it as the “kernel” of the OS: fast, safe, and predictable.<br> — <strong>gRPC</strong> gave me native service interactions, strongly typed contracts, and blazing-fast communication — all loosely coupled via dependency injection.<br> — A <strong>custom Redis Pub/Sub Event Bus</strong> became the nervous system. It allowed for decoupled communication across events, enabling real-time interactions.<br> — <strong>WebSockets</strong> provided a persistent two-way connection for live updates — like a window manager pushing system state to the user, and receiving user signals back into the Event Bus.<br> — <strong>Multithreaded WebAssembly (WASM) + WebGPU</strong> gave me low-level access to the user’s GPU, enabling compute and graphics acceleration directly in the browser — portable and sandboxed.<br> — <strong>Vite</strong> gave me a hot-reloadable, modular frontend with native support for TypeScript, WASM, and modules — a live participant in the system, tightly integrated with the Directed Acyclic Graph (DAG).</p><h3>That Thrilling Moment</h3><p>Watching the logs as the system booted — successfully, with all its components live — gave me an incredible thrill. It felt like building a working prototype for an old-school experiment. I didn’t stop there.</p><p>I went on to bootstrap a <strong>decentralized AI solution</strong> into it.</p><h3>The Bigger Picture</h3><p>This is a <strong>Directed Acyclic Graph (DAG)</strong> model, orchestrated over a <strong>Device-Agnostic Operating Substrate Layer</strong> that can tap into the power of <strong>Global</strong> compute.</p><p>If you’re inquisitive enough, smart enough, or just curious, you might already understand what I’m trying to achieve.</p><h3>Introducing INOS: The Internet Native Operating System</h3><p>The <strong>Internet Native Operating System (INOS)</strong> is still a work in progress. And I’ve come to realise this isn’t something I can — or should — build alone.</p><p>It’s open-sourced and waiting for your curiosity (along with more of my thoughts &amp; documentation) here: <em>[</em><a href="https://github.com/nmxmxh/master-ovasabi"><em>LINK</em></a><em>]</em></p><p>I’d love your help in building for the future. Let’s create the systems of tomorrow — together.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2c3261fa491b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[One for Accessibility, All for Accessibility]]></title>
            <link>https://nmxmxh.medium.com/one-for-accessibility-all-for-accessibility-0bb1c49903e9?source=rss-1ff1d334efdb------2</link>
            <guid isPermaLink="false">https://medium.com/p/0bb1c49903e9</guid>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[front-end-development]]></category>
            <category><![CDATA[accessibility]]></category>
            <category><![CDATA[user-experience]]></category>
            <category><![CDATA[inclusive-design]]></category>
            <dc:creator><![CDATA[Nobert Momoh]]></dc:creator>
            <pubDate>Thu, 08 May 2025 17:50:07 GMT</pubDate>
            <atom:updated>2025-05-08T17:50:07.367Z</atom:updated>
            <content:encoded><![CDATA[<blockquote>Building for one is building for all. Accessibility is not just a checklist — it’s a mindset that leads to better, more inclusive, and more successful digital experiences for everyone.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6QSjkkyR4xElnOOqj9vkdw.png" /></figure><p>When building frontend applications, I’ve always found it more meaningful to design for one person, not for “scale.” But who is this one person? Are they able to fully experience what I’ve built? Where are they from? Do they have a disability? Am I communicating in a language they understand? Can they access my assets and media, even with network limitations? These are the questions that shape my approach to accessibility.</p><h3>What Is Web Accessibility?</h3><p>Web accessibility means making your applications usable for people with disabilities — but it’s so much more. True accessibility is about empathy: creating digital experiences that everyone can enjoy, regardless of ability, context, or circumstance.</p><p>People may face a variety of challenges when using the web, including:</p><ul><li>Mobility and physical disabilities</li><li>Cognitive and neurological differences</li><li>Visual impairments</li><li>Hearing impairments</li></ul><p>Without a solid understanding of accessibility principles and semantic HTML, it’s easy for JavaScript and CSS to inadvertently create barriers. Overly complex interfaces, “div soup,” and non-semantic markup can make your site unusable for many.</p><h3>How People Access the Web</h3><p>There are many ways people interact with the web, including:</p><ul><li>Keyboard only navigation</li><li>Head wands and mouth sticks</li><li>Single-switch (one-button) devices</li><li>Screen readers (software that converts text to speech)</li></ul><p>Each of these methods requires thoughtful design and development to ensure a smooth experience.</p><h3>The Curb Cut Effect</h3><p>The “curb cut effect” describes how accessibility features designed for people with disabilities end up benefiting everyone. Curb cuts in sidewalks help not just wheelchair users, but also parents with strollers, travelers with luggage, and delivery workers. Similarly, accessible websites are easier to use for everyone — including people on mobile devices, with slow connections, or in noisy environments.</p><p>Accessible design also improves SEO: semantic HTML helps search engines understand your content, and features like alt text make your site more discoverable and usable.</p><h3>Accessibility Standards</h3><p>The <a href="https://www.w3.org/WAI/standards-guidelines/wcag/">Web Content Accessibility Guidelines (WCAG)</a> are the gold standard for web accessibility. They provide clear, actionable criteria for making your site accessible. WebAIM offers excellent checklists and resources to help you meet these standards.</p><h3>Screen Readers</h3><p>Screen readers, such as <a href="https://www.nvaccess.org/">NVDA</a> and <a href="https://www.freedomscientific.com/products/software/jaws/">JAWS</a>, convert digital text into synthesised speech. To support them:</p><ul><li>Use descriptive <a href="https://webaim.org/techniques/alttext/">alt text</a> for images.</li><li>Provide captions and transcripts for audio and video.</li><li>Use semantic HTML so screen readers can interpret your content structure.</li><li>AI is making screen readers smarter, but clear markup is still essential.</li></ul><h3>Accessible HTML</h3><ul><li><strong>Use semantic HTML:</strong> Avoid “div soup.” Elements like &lt;header&gt;, &lt;nav&gt;, &lt;main&gt;, &lt;aside&gt;, and &lt;footer&gt; provide structure for both users and search engines.</li><li><strong>Only one &lt;h1&gt; per page:</strong> This helps with both accessibility and SEO.</li><li><strong>Label form fields properly:</strong> Use &lt;label&gt; elements, not &lt;p&gt; tags, and ensure each label is associated with its input.</li><li><strong>ARIA attributes:</strong> Use <a href="https://developer.mozilla.org/en-US/docs/Web/Accessibility/ARIA/Attributes/aria-label">ARIA labels</a> and roles to enhance accessibility, but don’t use them as a substitute for semantic HTML.</li><li><strong>Tab index:</strong> Manage keyboard navigation with tabindex.</li><li><strong>Keyboard events:</strong> Use onKeyUp and similar events to support keyboard users.</li></ul><h3>ARIA: Accessible Rich Internet Applications</h3><p><a href="https://www.w3.org/WAI/standards-guidelines/aria/">ARIA</a> helps make dynamic content and custom UI components accessible. Use ARIA roles, states, and properties to describe elements to assistive technologies. For example, use aria-checked for custom checkboxes, and aria-describedby to provide additional context.</p><p><strong>Note:</strong> Overusing ARIA can cause confusion. Prefer native HTML elements whenever possible.</p><h3>Forms Management</h3><ul><li><strong>Support keyboard-only users:</strong> Provide clear focus indicators and logical tab order.</li><li><strong>Don’t change pages unexpectedly:</strong> Sudden changes can disorient users.</li><li><strong>Keyboard shortcuts:</strong> Offer shortcuts, but make them discoverable and avoid conflicts.</li></ul><h3>Skip Links</h3><p>Skip links let users jump directly to main content, bypassing navigation. Place a visually hidden menu at the top of your site that becomes visible on focus. <a href="https://webaim.org/techniques/skipnav/">Learn more about skip links.</a></p><h3>Tab Navigation</h3><ul><li><strong>Tabbable elements:</strong> Only interactive elements (links, buttons, form fields) should be focusable.</li><li><strong>Arrow keys:</strong> Use for vertical navigation in menus.</li><li><strong>Focus management:</strong> Use the <a href="https://developer.mozilla.org/en-US/docs/Web/API/Document/activeElement">DOM API’s `activeElement`</a> to manage focus, especially in modals (focus trapping).</li></ul><h3>Visual Considerations</h3><ul><li><strong>Escape closes modals:</strong> Users expect this.</li><li><strong>Color contrast: </strong>Use tools like the <a href="https://webaim.org/resources/contrastchecker/">WebAIM Contrast Checker</a> to ensure readability.</li><li><strong>Don’t rely on color alone:</strong> Use icons, text, or patterns to convey meaning.</li><li>Simulate color blindness: Tools like <a href="https://addons.mozilla.org/en-US/firefox/addon/nocoffee/">NoCoffee for Firefox</a> help you test accessibility.</li></ul><h3>Language and Markup</h3><ul><li><strong>Set the page language:</strong> Use the lang attribute on &lt;html&gt;.</li><li><strong>Validate your markup: </strong>Use the <a href="https://validator.w3.org/">W3C Markup Validation Service</a>.</li><li><strong>Internationalization:</strong> Use libraries like <a href="https://formatjs.github.io/">FormatJS</a> for i18n.</li></ul><h3>Motion and Color Scheme Preferences</h3><ul><li>Reduced motion: Respect users’ <a href="https://developer.mozilla.org/en-US/docs/Web/CSS/@media/prefers-reduced-motion">prefers-reduced-motion</a> settings.</li><li>Color scheme: Support <a href="https://developer.mozilla.org/en-US/docs/Web/CSS/@media/prefers-color-scheme">prefers-color-scheme</a> for light/dark mode.</li></ul><h3>Tooling</h3><ul><li><strong>Linters:</strong><br>- <a href="https://github.com/jsx-eslint/eslint-plugin-jsx-a11y">eslint-plugin-jsx-a11y</a><br>- <a href="https://github.com/mgechev/codelyzer">angular codelyzer<br></a>- <a href="https://github.com/vue-a11y/eslint-plugin-vuejs-accessibility">eslint-plugin-vuejs-accessibility</a></li><li><strong>Design Systems:</strong><br>- <a href="https://react-spectrum.adobe.com/react-spectrum/">Adobe React Spectrum</a><br>- <a href="https://m3.material.io/">Google Material Design</a></li><li><strong>Developer Tools: </strong><br>- <a href="https://www.deque.com/axe/devtools/">Deque axe DevTools</a><br>- <a href="https://developers.google.com/web/tools/lighthouse">Google Lighthouse</a></li></ul><h3>Resources</h3><ul><li><a href="https://www.w3.org/WAI/standards-guidelines/wcag/">Web Content Accessibility Guidelines (WCAG)</a></li><li><a href="https://webaim.org/">WebAIM</a></li><li><a href="https://www.microsoft.com/design/inclusive/">Microsoft Inclusive Design</a></li><li><a href="https://accessibility.day/">Global Accessibility Awareness Day</a></li><li><a href="https://www.smashingmagazine.com/2021/03/accessibility-javascript-applications/">Accessibility in JavaScript Applications</a></li><li><a href="https://web.dev/accessibility/">Start Building Accessible Web Applications Today</a></li><li><a href="https://www.a11yproject.com/checklist/">Accessibility Tips &amp; Tricks</a></li></ul><p><strong>Special Note:</strong></p><ul><li>Anchors and buttons must have text content.</li><li>Lists should be meaningful and properly structured.</li></ul><p><strong>Remember:</strong></p><blockquote>Building for one is building for all. Accessibility is not just a checklist — it’s a mindset that leads to better, more inclusive, and more successful digital experiences for everyone.</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0bb1c49903e9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Beyond the Average: Giving AI the Context It Deserves]]></title>
            <link>https://nmxmxh.medium.com/beyond-the-average-giving-ai-the-context-it-deserves-7bf88818de60?source=rss-1ff1d334efdb------2</link>
            <guid isPermaLink="false">https://medium.com/p/7bf88818de60</guid>
            <category><![CDATA[machine-learning]]></category>
            <category><![CDATA[software-architecture]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[knowledge-graph]]></category>
            <category><![CDATA[engineering-productivity]]></category>
            <dc:creator><![CDATA[Nobert Momoh]]></dc:creator>
            <pubDate>Tue, 06 May 2025 12:08:39 GMT</pubDate>
            <atom:updated>2025-05-06T12:25:08.151Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Go7EGivHiqbZqWBqxnBGxQ.png" /></figure><blockquote>“AI is only as good as the context and examples you give it. With the right knowledge graph and scope, it can become your team’s most productive expert.”</blockquote><h3>Introduction</h3><p>Artificial Intelligence (AI) is rapidly transforming how we build, operate, and reason about complex systems. Yet, despite its scale and speed, AI often struggles with context, best practices, and architectural nuance. This article explores the fundamental contrast between AI and human intelligence, the problem of “median bias” in AI, and how context, knowledge graphs, and curated references can unlock AI’s true productivity for technical teams.</p><h3>Human Intelligence vs. AI Intelligence: The Context Gap</h3><h4>Human Intelligence</h4><ul><li><strong>Starts from nothing</strong>: Humans are born with no knowledge and slowly build intelligence through experience, mentorship, and exposure to best practices.</li><li><strong>Contextual learning</strong>: Human experts develop intuition and judgment by learning from high-quality examples, mistakes, and domain-specific mentorship.</li><li><strong>Bias toward excellence</strong>: With the right environment, humans can learn from the best and avoid common pitfalls, leading to mastery and innovation.</li></ul><h4>AI Intelligence</h4><ul><li><strong>Starts with scale</strong>: AI models (like LLMs) are trained on vast, diverse datasets, absorbing the “average” of human knowledge and code.</li><li><strong>Median bias:</strong> Most training data represents the median, not the best. AI is excellent at producing “average” solutions, but struggles with optimal architecture, best practices, or domain-specific nuance.</li><li><strong>Lacks context:</strong> Without explicit context, AI cannot distinguish between good, bad, or outdated patterns. It has no “intuition” for what works best in your environment.</li></ul><h3>The Problem: AI’s Median Bias in Practice</h3><p>AI can generate code, docs, and designs at scale, but…<br> — It often proposes “safe” or “common” solutions, not the best ones for your system.<br> — It may miss architectural patterns, compliance requirements, or performance optimisations that an expert would know.<br> — It cannot infer your team’s standards, legacy constraints, or business goals without explicit context.</p><p><strong>Example:</strong></p><p>Ask an LLM to design a microservices architecture. It will produce a generic, textbook answer — ignoring your team’s preferred stack, deployment model, or integration patterns.</p><h3>The Solution: Context, Knowledge Graphs, and Curated Scope</h3><h4>1. Provide Explicit Context</h4><ul><li>Give the AI agent detailed information about your system, goals, and constraints.</li><li>Use architectural diagrams, service registries, and documentation as input.</li><li>The more context, the more relevant and high-quality the AI’s output.</li></ul><h4>2. Leverage Knowledge Graphs</h4><ul><li>A knowledge graph encodes the relationships, patterns, and best practices of your system.</li><li>It acts as a “memory” for the AI, allowing it to reason about dependencies, compliance, and impact.</li><li>With a knowledge graph, AI can:<br> — Suggest changes that fit your architecture<br> — Warn about breaking changes or anti-patterns<br> — Recommend proven solutions from your own history</li></ul><h4>3. Limit Scope to Good and Tenured References</h4><ul><li>Curate the data and examples the AI can access.</li><li>Prioritize references from your best engineers, successful projects, and trusted sources.</li><li>Exclude outdated, low-quality, or irrelevant patterns.</li><li>This “scope limiting” turns AI from a generic assistant into a domain expert.</li></ul><h3>Maximising AI Productivity: A Practical Approach</h3><ol><li>Build and maintain a system knowledge graph.</li><li>Integrate the knowledge graph with your AI agents.</li><li>Feed the AI high-quality, context-rich prompts.</li><li>Continuously update the graph with new patterns, migrations, and lessons learned.</li><li>Review and curate the AI’s outputs, feeding back improvements.</li></ol><h3>Conclusion</h3><p>AI’s power comes from scale, but its productivity and architectural quality depend on context. By bridging the gap between median-trained models and your team’s best practices using knowledge graphs, curated references, and explicit context you can turn AI from a generic code generator into a true engineering accelerator.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7bf88818de60" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Creating Master: An Ovasabi Case Study]]></title>
            <link>https://nmxmxh.medium.com/creating-master-an-ovasabi-case-study-8f43432cc5f2?source=rss-1ff1d334efdb------2</link>
            <guid isPermaLink="false">https://medium.com/p/8f43432cc5f2</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[referral-marketing]]></category>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[community]]></category>
            <category><![CDATA[go]]></category>
            <dc:creator><![CDATA[Nobert Momoh]]></dc:creator>
            <pubDate>Mon, 05 May 2025 14:02:30 GMT</pubDate>
            <atom:updated>2025-05-05T14:02:30.288Z</atom:updated>
            <content:encoded><![CDATA[<p><em>By Nobert Momoh, Creative Director, </em><a href="https://www.linkedin.com/company/ovasabi-studios"><em>Ovasabi Studios</em></a></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Wkr7lT_RnQG-6v5pm5Yf1Q.jpeg" /></figure><blockquote><em>“To create an apple pie from scratch, you must first create the universe.”<br> — Carl Sagan</em></blockquote><p>I’ve long been fascinated by the Renaissance — a cultural rebirth from the 14th to the 17th century that transformed art, science, and society. It wasn’t just a revival of classical forms; it was the moment when the artist transcended the craftsman’s role to become a true intellectual creator.</p><p>One of my heroes from that era is <strong>Lorenzo de’ Medici</strong>. If I could sit down with anyone, living or dead, Lorenzo would be at the top of the list (with Walt Disney close behind). As Florence’s de facto ruler, Lorenzo’s visionary patronage gave us Michelangelo’s majesty, Leonardo’s genius, and Botticelli’s myths. He didn’t merely bankroll projects — he cultivated a vibrant court of thinkers and artists, making Florence the beating heart of Renaissance innovation.</p><p>Inspired by this legacy, I founded <strong>Ovasabi Studios</strong> as a rebellion against the anti‑intellectual currents in my society. Our mission is to create a hub where creators, visionaries, and entrepreneurs can collaborate, innovate, and thrive. If you’d like to keep up with our journey, connect with us on LinkedIn: <a href="https://www.linkedin.com/company/ovasabi-studios">Ovasabi Studios</a>.</p><h3>From Carl Sagan to “Master”</h3><p>In Sagan’s words, making an apple pie “from scratch” means understanding every cosmic process that forged its ingredients. At Ovasabi, we take that to heart: building modern software “from scratch” means mastering every layer — from data models to deployment pipelines.</p><p>And so <strong>Master</strong> was born.</p><h3>What Is Master?</h3><p><strong>Master</strong> is a next‑generation backend platform designed to power modern digital products with agility and intelligence. At its core lies <strong>Amadeus</strong>, a self‑documenting knowledge graph that continuously maps every service, data model, and relationship in your system. With Amadeus, both developers and AI agents have real‑time, programmatic access to your architecture — enabling:</p><ul><li><strong>Instant documentation</strong>: Always‑up‑to‑date API and data model specs</li><li><strong>Impact analysis</strong>: Trace downstream effects of any change</li><li><strong>Smarter automation</strong>: Let AI agents configure, deploy, and evolve your platform</li></ul><h3>Modular Services, Unified Experience</h3><p>Master offers a rich suite of services you can pick and choose:</p><ul><li><strong>Asset Management</strong>: Real‑time optimization and streaming for images, videos, and 3D models</li><li><strong>Authentication &amp; Notifications</strong>: Secure sign‑on flows and multi‑channel alerts</li><li><strong>Finance &amp; Quotes</strong>: Billing, invoicing, and dynamic quoting engines</li><li><strong>Campaign Orchestration</strong>: Design, schedule, and monitor marketing workflows</li><li><strong>Internationalization</strong>: Babel service for multilingual content and location‑based pricing</li></ul><p>Every service registers with a central Provider/Dependency‑Injection container — ensuring consistency, discoverability, and frictionless extensibility.</p><h3>Nexus: Workflow Orchestration</h3><p>Our <strong>Nexus</strong> engine lets you compose complex business processes as declarative graphs. Trigger events, call services, branch logic — and let Nexus handle retries, monitoring, and error recovery.</p><h3>Community at the Core</h3><p>Master’s referral system isn’t an afterthought; it’s a core philosophy. By rewarding users for engagement, we’re building a self‑sustaining, collaborative community. And with Amadeus powering insights, everyone — human or AI — can learn, grow, and build together.</p><h3>Next Steps</h3><p>More than code, <strong>Master</strong> feels like a work of art. Over the coming months, we’ll be refining its APIs, expanding our service catalog, and nurturing the Ovasabi community into a true Renaissance of digital innovation.</p><p>Stay tuned — and don’t forget to follow us on LinkedIn for the latest updates: <a href="https://www.linkedin.com/company/ovasabi-studios">Ovasabi Studios</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8f43432cc5f2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[I created my dream backend with Go as React dev with Cursor; and asked it how fast could it go.]]></title>
            <link>https://nmxmxh.medium.com/i-created-my-dream-backend-with-go-as-react-dev-with-cursor-and-asked-it-how-fast-could-it-go-8f91f6f235ad?source=rss-1ff1d334efdb------2</link>
            <guid isPermaLink="false">https://medium.com/p/8f91f6f235ad</guid>
            <category><![CDATA[styled-components]]></category>
            <category><![CDATA[nextjs]]></category>
            <category><![CDATA[react]]></category>
            <category><![CDATA[go]]></category>
            <category><![CDATA[grpc]]></category>
            <dc:creator><![CDATA[Nobert Momoh]]></dc:creator>
            <pubDate>Tue, 29 Apr 2025 20:32:12 GMT</pubDate>
            <atom:updated>2025-04-29T20:32:12.591Z</atom:updated>
            <content:encoded><![CDATA[<p>This was it’s response.</p><h3>Great question! Let’s break down the performance and scalability of this information channel approach, and how you can make it highly concurrent and scalable for global campaigns.</h3><h3>1. How Fast Could This Go?</h3><h3>A. Theoretical Throughput</h3><ul><li>Concurrency:</li><li>Each message send (to a user) is independent.</li><li>You can process/broadcast to thousands or millions of users in parallel, limited only by your infrastructure and implementation.</li><li>Network/IO Bound:</li><li>The main bottlenecks are:</li><li>Fetching user data (location/language)</li><li>Fetching translations from i18n</li><li>Sending messages (e.g., to notification/email/push systems)</li><li>All of these can be parallelized.</li></ul><h3>B. Real-World Numbers</h3><ul><li>With a well-implemented Go service (using goroutines, worker pools, and batching), you can easily handle:</li><li>10,000+ messages per second on a single modern server, if the downstream systems (e.g., notification providers, DB, Redis) can keep up.</li><li>Horizontal scaling:</li><li>You can run multiple instances of your broadcast service, each processing a subset of users/messages.</li></ul><h3>2. How to Make It Concurrent</h3><h3>A. Use Goroutines and Worker Pools</h3><ul><li>For each user, spawn a goroutine (or use a worker pool to avoid oversubscription).</li><li>Use channels to queue work and collect results/errors.</li></ul><h3>B. Batch DB and i18n Calls</h3><ul><li>Fetch user data and translations in batches where possible to reduce round-trips.</li></ul><h3>C. Asynchronous Delivery</h3><ul><li>For very large campaigns, consider:</li><li>Writing jobs to a queue (e.g., Redis, RabbitMQ, Kafka)</li><li>Having multiple workers/processes consume and send messages concurrently</li></ul><h3>D. Rate Limiting and Backoff</h3><ul><li>Implement rate limiting to avoid overwhelming downstream services (e.g., push/email providers).</li><li>Use exponential backoff/retries for transient errors.</li></ul><h3>3. How Much Scale Could It Handle?</h3><h3>A. Database and Service Limits</h3><ul><li>Postgres:</li><li>Can handle thousands of concurrent reads/writes if tuned and indexed properly.</li><li>For massive scale, consider read replicas or sharding.</li><li>Redis:</li><li>Extremely fast for caching and queueing; can handle hundreds of thousands of ops/sec.</li><li>Go Services:</li><li>Go’s goroutines are lightweight; you can run tens of thousands concurrently on a single server.</li></ul><h3>B. Horizontal Scaling</h3><ul><li>Stateless services:</li><li>Your broadcast workers can be scaled horizontally (just run more containers/VMs).</li><li>Queue-based architecture:</li><li>Use a message queue to distribute work across many workers for virtually unlimited scale.</li></ul><h3>C. Real-World Examples</h3><ul><li>Systems like this (with proper batching, queuing, and horizontal scaling) can handle millions of messages per hour.</li><li>For reference, large SaaS platforms and notification systems use similar patterns to reach millions of users in minutes.</li></ul><h3>4. Bottlenecks &amp; Solutions</h3><p>| Bottleneck | Solution |</p><p>| — — — — — — — — — — | — — — — — — — — — — — — — — — — — — — — — — — -|</p><p>| DB read/write | Batch queries, use replicas, index properly |</p><p>| i18n lookups | Cache translations in memory or Redis |</p><p>| Message delivery | Use async queues, batch sends, retry/backoff |</p><p>| Network/IO | Use connection pools, parallelism |</p><p>| Downstream limits | Rate limit, monitor, and autoscale workers |</p><h3>5. Example: Go Worker Pool for Broadcast</h3><pre>var wg sync.WaitGroup<br>userCh := make(chan User, 1000) // buffered channel<br><br>// Start N workers<br>for i := 0; i &lt; numWorkers; i++ {<br>    wg.Add(1)<br>    go func() {<br>        defer wg.Done()<br>        for user := range userCh {<br>            // Fetch translation, send message, etc.<br>        }<br>    }()<br>}<br><br>// Enqueue users<br>for _, user := range users {<br>    userCh &lt;- user<br>}<br>close(userCh)<br>wg.Wait()</pre><ul><li>This pattern can easily scale to tens of thousands of concurrent sends.</li></ul><h3>6. Summary Table</h3><p>| Scale Level | How to Achieve |</p><p>| — — — — — — — — — — -| — — — — — — — — — — — — — — — — — — — — — — — — — |</p><p>| 10,000/sec | Goroutines, batching, async sends |</p><p>| 100,000+/sec | Add horizontal scaling, distributed queues |</p><p>| Millions/hour | Use cloud infra, autoscale, global distribution |</p><h3>Conclusion</h3><ul><li>Your architecture is highly scalable and concurrent by design.</li><li>With proper Go concurrency, batching, and queueing, you can reach global scale.</li><li>The main limits will be your infrastructure and any external providers (e.g., SMS, email, push).</li></ul><p>If you want a sample implementation of a concurrent broadcast worker, or advice on tuning for even larger scale, just ask!</p><p>-</p><p>I sit, now scared of the implications.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8f91f6f235ad" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[To New Beginnings.]]></title>
            <link>https://nmxmxh.medium.com/to-new-beginnings-e0b74adb3b41?source=rss-1ff1d334efdb------2</link>
            <guid isPermaLink="false">https://medium.com/p/e0b74adb3b41</guid>
            <category><![CDATA[failure]]></category>
            <category><![CDATA[introduction]]></category>
            <category><![CDATA[insecurity]]></category>
            <category><![CDATA[beginning]]></category>
            <dc:creator><![CDATA[Nobert Momoh]]></dc:creator>
            <pubDate>Sun, 17 Jan 2021 11:57:01 GMT</pubDate>
            <atom:updated>2021-01-17T11:57:01.990Z</atom:updated>
            <content:encoded><![CDATA[<p>Hello,</p><p>It’s really been a while since I’ve offered my words on the altar of human consumption, parted away with a bit of myself to be judged, critiqued and assessed. Still, while starting this introduction, I still notice the anxiety and insecurity hidden in my prose, a product of my constant need for perfection, praise or validation for my efforts, and in this moment, I accept them.</p><p>This might come as a surprise (well, to me), but to be good at something, you should be willing to be bad at it first. This should be common sense but I’ve rarely committed to the consistency and perseverance required to fail, fail and fail some more till you’re a little better at not failing.</p><p>So I’ll keep this short, thank you for reading my words, dear reader. There’s a lot more to come, a lot of places we will venture into and a lot of failures we will encounter but at the end of it all, the true treasures are the friends we find during this journey.</p><p>I’m kidding, I’m kidding. It’s a joke.</p><p>My name is Momoh Nobert and finally, I’m writing for human consumption again.</p><p>You’re welcome.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e0b74adb3b41" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>