<?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 Wesley Huber on Medium]]></title>
        <description><![CDATA[Stories by Wesley Huber on Medium]]></description>
        <link>https://medium.com/@wesleybaxterhuber?source=rss-cb71643b03ea------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*GgsVuxdI3Lf-ZqFay3iH_Q.jpeg</url>
            <title>Stories by Wesley Huber on Medium</title>
            <link>https://medium.com/@wesleybaxterhuber?source=rss-cb71643b03ea------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 21:08:01 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@wesleybaxterhuber/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[Zero Knowledge Proofs: How to Win Every “Trust Me Bro” Argument With Math]]></title>
            <link>https://wesleybaxterhuber.medium.com/zero-knowledge-proofs-how-to-win-every-trust-me-bro-argument-with-math-819ba1bc96b2?source=rss-cb71643b03ea------2</link>
            <guid isPermaLink="false">https://medium.com/p/819ba1bc96b2</guid>
            <category><![CDATA[zero-knowledge-proofs]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[cryptography]]></category>
            <category><![CDATA[crypto]]></category>
            <category><![CDATA[blockchain]]></category>
            <dc:creator><![CDATA[Wesley Huber]]></dc:creator>
            <pubDate>Mon, 27 Jul 2026 21:58:56 GMT</pubDate>
            <atom:updated>2026-07-28T04:21:19.625Z</atom:updated>
            <content:encoded><![CDATA[<h3><strong>A tutorial where you prove things without revealing things, and yes, the math actually maths.</strong></h3><p>Here’s something the internet doesn’t want you to know: <strong>you overshare every single time you prove something.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/920/1*ojowa4vBK549LRwFYY8_MA.png" /></figure><p>Prove you’re over 21 at a bar? You hand over a card with your name, your address, your height, and your terrible 2019 haircut. Prove your income to a landlord? Here’s every transaction I’ve made since college, please don’t judge the 3am food delivery.</p><p>We built the entire digital world on a verification model that boils down to “<strong>here’s everything, trust me bro.”</strong></p><p>Not anymore. There’s a branch of cryptography that lets you prove a statement is true while revealing <em>*nothing else*</em>. It sounds fake. It’s called a <strong>zero knowledge proof</strong>, and by the end of this article you’ll understand one well enough to check it with Python. Then we’ll look at <a href="https://midnight.network/"><strong>Midnight</strong></a>, a blockchain that turned this party trick into a developer platform.</p><p>Let’s go. 🚀</p><p><strong>🪪 The Trust Me Bro Problem</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/686/1*EdD87ht_ifiX3V-Xs2ExMQ.png" /></figure><p>Every verification system you use today works by <strong>disclosure</strong>. You prove things by showing the underlying data:</p><p>- Prove your age ➡️ show your whole ID</p><p>- Prove you can pay ➡️ show your bank statements</p><p>- Prove you’re a real user ➡️ solve a CAPTCHA and sacrifice your data to the algorithm gods</p><p>The data doesn’t just get <em>seen</em>. It gets <strong>stored</strong>, and eventually it gets <strong>breached</strong>, and then a guy named xX_darkweb_Xx is selling your identity for the price of a burrito.</p><p>The verifier never needed the data. They needed <strong>one bit of information</strong>: true or false. Everything else was collateral damage.</p><p><strong>In short:</strong> we’ve been answering yes or no questions with our entire life story.</p><h3><strong>🕵️ The Party Trick That Started It All</strong></h3><p>Zero knowledge proofs let a <strong>prover</strong> convince a <strong>verifier </strong>that a statement is true without revealing <em>why</em> it’s true.</p><p>The classic example is Where’s Waldo. Say I claim I found Waldo on the page and you don’t believe me (fair, you’ve seen my code reviews). I could point at him, but then I’ve revealed the answer and ruined the puzzle.</p><p>Instead, I take a giant piece of cardboard, way bigger than the book, cut a tiny Waldo shaped hole in it, and slide the page underneath so only Waldo shows through the hole.</p><p>You see Waldo. ✅ You learn I found him. ✅ You learn <strong>**absolutely nothing**</strong> about where he is on the page, because the cardboard hides all of it. ✅</p><p>That’s a zero knowledge proof. Every real ZK system, no matter how much math it’s wearing, is doing the cardboard trick.</p><p>The formal version has three properties, and knowing these three words will let you nod convincingly in any crypto conversation:</p><p><strong>Completeness</strong>: if the statement is true, an honest prover convinces the verifier. The demo works.</p><p><strong>Soundness</strong>: if the statement is false, no amount of galaxy brain trickery convinces the verifier. You can’t fake it.</p><p><strong>Zero knowledge</strong>: the verifier learns the statement is true and <em>nothing else</em>. The cardboard stays intact.</p><p>Good job, concept unlocked! 💪 Now let’s earn the “I understand the math” badge.</p><h3><strong>🧮 The Math (This Is Where You Start Feeling Smart)</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/666/1*Jn-1lFHk2G-hPN_ENFsdzA.png" /></figure><p>Don’t close the tab. The math is genuinely simple, it’s just wearing a trench coat.</p><p>Everything runs on <strong>clock math</strong>, which mathematicians call modular arithmetic to keep it gatekept. On a 12 hour clock, 9 + 4 = 1. You wrap around. We write that as 9 + 4 ≡ 1 (mod 12). That’s it. That’s the whole trench coat.</p><p>Here’s the fun part. Pick a prime <strong>p</strong> and a number <strong>g</strong>. Computing y = g^x mod p is easy, your laptop does it instantly. But going <em>backwards</em>, finding x when you only know y, is called the <strong>discrete logarithm problem</strong>, and for big numbers it’s computationally hopeless. Multiplying is easy, unmultiplying is a research career.</p><p>So: <strong>x can be a secret, and y can be public.</strong> Think of x as a private key and y as a public key. Now I’ll prove I know x without ever showing it, using a protocol cryptographers named after Claus Schnorr (cryptographers name everything after themselves, it’s their love language).</p><p>The protocol is a three message dance between Peggy (<strong>P</strong>rover) and Victor (<strong>V</strong>erifier):</p><p>1. <strong>Commit.</strong> Peggy picks a random number k and sends r = g^k mod p. This locks her in, like pushing a commit before the standup so everyone knows you did <em>something</em>.</p><p>2. <strong>Challenge.</strong> Victor sends back a random number c. This is the crucial part: Peggy couldn’t predict it.</p><p>3. <strong>Respond.</strong> Peggy sends s = k + c · x. Her secret x is in there, but it’s blended with the random k, like a password inside a hash. Unrecoverable.</p><p>Victor then checks one equation:</p><p><strong>g^s ≡ r · y^c (mod p)</strong></p><p>If Peggy really knows x, the algebra works out perfectly every time (completeness). If she’s bluffing, she’d need to answer a random challenge she couldn’t predict, and she gets caught (soundness). And Victor only ever sees r, c, and s, which are statistically just noise (zero knowledge).</p><p>Let’s run it with real numbers, because “trust me bro” would be ironic here.</p><p>- Setup: p = 23, g = 5. Peggy’s secret: x = 6. Public value: y = ⁵⁶ mod 23 = <strong>8</strong></p><p>- Commit: Peggy picks k = 3, sends r = ⁵³ mod 23 = <strong>10</strong></p><p>- Challenge: Victor sends c = <strong>4</strong></p><p>- Respond: s = 3 + 4 · 6 = <strong>27</strong></p><p>- Victor checks: is ⁵²⁷ mod 23 equal to 10 · ⁸⁴ mod 23?</p><p>Open a Python shell and verify it yourself:</p><pre>&gt;&gt;&gt; pow(5, 27, 23)<br>20<br>&gt;&gt;&gt; (10 * pow(8, 4, 23)) % 23<br>20</pre><p><strong>20 == 20</strong> The math is mathing. Victor is convinced Peggy knows x, and if you scroll back, x = 6 never appeared in anything Victor saw. 🤯</p><p>If you’ve made it this far you are a superstar ⭐️ and you now understand more actual cryptography than 99% of the people posting candle charts. Few understand this. You do.</p><h3><strong>🚀 From Party Trick to zkSNARK</strong></h3><p>Our dance has two problems for the real world. Victor had to be online to throw challenges, and we only proved one tiny algebra fact.</p><p>Modern systems fix both:</p><p><strong>Kill the back and forth.</strong> Instead of Victor picking the challenge, Peggy generates it by hashing her own commitment (the Fiat Shamir transform). The hash function becomes an incorruptible robot Victor. Now the proof is a single message anyone can verify, anytime.</p><p><strong>Prove anything.</strong> Any computation can be compiled into a giant system of equations like our little one. Your program becomes a circuit, and you prove “I ran this code correctly on secret inputs” the same way Peggy proved she knew x.</p><p>Bundle that up and you get a <strong>zkSNARK</strong>: a proof that’s <em>succinct</em> (a few hundred bytes even if the computation was enormous) and verifiable in milliseconds.</p><p>“It works on my machine” finally has a fix. With a zkSNARK it provably works on <strong>**every**</strong> machine, and you don’t even have to show anyone the machine. 🐳</p><p><strong>In short:</strong> you can now hand someone a receipt that proves a computation happened correctly, without showing the inputs. Which raises a question with billion dollar implications: what happens when you put that on a blockchain?</p><h3><strong>🌙 Enter Midnight</strong></h3><p>Public blockchains have a privacy model best described as</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/934/1*En-dXAN1Osk775LzG6fOow.png" /></figure><p>Radical transparency is great for verifying money and <em>terrible</em> for literally everything else. No company is putting payroll on a public ledger. No hospital is putting your records there. Transparency was blockchain’s superpower and its adoption ceiling at the same time.</p><p><strong>Midnight</strong> is a blockchain built to fix exactly this. It’s a partner chain in the Cardano ecosystem designed around <strong>data protection</strong>: smart contracts where zero knowledge proofs are not an exotic bolt on, they’re the default way state gets updated. Your data stays on your device. The chain sees proofs.</p><p>The part I love as a developer: you don’t write moon math. You write <strong>Compact</strong>, a TypeScript flavored language, and the toolchain compiles your logic into ZK circuits. Here’s the vibe (simplified for the article, real syntax lives in their docs):</p><pre>// the witness lives on YOUR device, never on chain<br>witness birthYear(): Uint&lt;16&gt;;<br>export circuit proveAdult(currentYear: Uint&lt;16&gt;): [] {<br>assert(currentYear - birthYear() &gt;= 18, &quot;must be an adult&quot;);<br>}</pre><p>The chain learns “this person is an adult: true.” Your birth year stays home. Peggy would be proud.</p><p>What does that unlock? The use cases write themselves:</p><ul><li>✅ <strong>Age and identity checks </strong>that don’t photocopy your life. Prove “over 18” or “licensed doctor” as one bit, not one dossier.</li><li>✅ <strong>Financial privacy with receipts.</strong> Prove solvency to a lender or an exchange without opening your books to the whole internet.</li><li>✅ <strong>Healthcare eligibility.</strong> Prove “vaccinated” or “matches this trial” while the medical record never leaves the building.</li><li>✅ <strong>Compliance without surveillance.</strong> Selective disclosure means a regulator with the right keys can audit you, while everyone else sees nothing. Regulated <em>*and*</em> private, which used to be a contradiction.</li><li>✅ <strong>Private voting and DAOs.</strong> Prove your vote counted without revealing it. Governance without vote buying.</li><li>✅ <strong>Enterprise logic on chain.</strong> Supply chains and business rules where the <em>*correctness*</em> is public but the trade secrets aren’t.</li></ul><p>Notice the pattern. Every one of these is a yes or no question we currently answer with a full data dump. ZK flips the default: <strong>**share the conclusion, keep the evidence.**</strong></p><h3><strong>🛠 Try It Yourself</strong></h3><p>The best part is you can actually build with this today:</p><p>1. <strong>Docs:</strong> <a href="https://docs.midnight.network">docs.midnight.network</a> has the full developer guide and the Compact language reference.</p><p>2. <strong>Write a circuit:</strong> the tutorial walks you through your first contract, and it genuinely feels like TypeScript, not like a math PhD hazing ritual.</p><p>3. <strong>Go deeper:</strong> the node itself is open source at <a href="https://github.com/midnightntwrk/midnight-node">github.com/midnightntwrk/midnight-node.</a> I’ve been contributing there lately and the codebase is a great read if you want to see how a privacy chain actually ships.</p><p>That’s it! We’re done yay 😁</p><h3><strong>🧾 The Recap</strong></h3><p>* Current verification = trust me bro + oversharing.</p><p>* Zero knowledge proofs = the Waldo cardboard: prove the claim, hide everything else.</p><p>* You personally verified a Schnorr proof with two lines of Python. Flex accordingly.</p><p>* zkSNARKs make it succinct and universal.</p><p>* Midnight makes it programmable in a TypeScript flavored language, with use cases that read like a list of everything Web2 got wrong about data.</p><p>The cryptographers spent forty years building the cardboard. The chains are finally cheap and the tooling finally speaks TypeScript. The only missing ingredient is developers who see the use case sitting in their own industry.</p><p>So: what can <em>*you*</em> prove?</p><p>If you build something with this, I genuinely want to hear about it. Drop a comment or find me on GitHub as wbaxterh.</p><p>Peace out,</p><p>Wes</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=819ba1bc96b2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[I’m Building Instagram, WhatsApp, and Netflix — For Skaters]]></title>
            <link>https://wesleybaxterhuber.medium.com/im-building-instagram-whatsapp-and-netflix-for-skaters-98ea9d2838e8?source=rss-cb71643b03ea------2</link>
            <guid isPermaLink="false">https://medium.com/p/98ea9d2838e8</guid>
            <dc:creator><![CDATA[Wesley Huber]]></dc:creator>
            <pubDate>Mon, 09 Feb 2026 00:00:37 GMT</pubDate>
            <atom:updated>2026-02-09T18:09:12.024Z</atom:updated>
            <content:encoded><![CDATA[<h3>Building in Public: Instagram, WhatsApp, &amp; Netflix. For Skaters</h3><p>How coding agents let one developer ship what used to take a team of 50</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*WNQFvM6sRKMxBXPk.png" /></figure><h3>The dirty secret of 2026</h3><p>Here’s something the tech industry doesn’t want you to know: <strong>the moat is gone.</strong></p><p>Building a social feed? That used to require a team of engineers, a $10M seed round, and 18 months of runway. Building real-time messaging? Add another $5M and a dedicated infrastructure team. Video streaming with adaptive bitrate? Better raise a Series A.</p><p>Not anymore.</p><p>I’m one guy with a coding agent, and I just shipped all three. For skateboarders.</p><h3>What I Actually Built</h3><p><strong>TrickBook</strong> started as a simple idea: let skaters track which tricks they’ve landed. A checkbox app. Version 1.0.</p><p>Then I kept going.</p><p><strong>The Feed</strong> — An Instagram-style social feed where riders share clips, react with 🤙 and ❤️, and discover new skaters. The algorithm prioritizes your “homies” (friends) and weighs engagement, recency, and completion rate. No ads. No shadowbanning. Just clips.</p><p><strong>DMs</strong> — WhatsApp-caliber messaging. Real-time delivery with WebSockets. Read receipts. Conversation threads. Plan sessions, share spots, talk shit about that kid who claims he landed a tre flip.</p><p><strong>The Couch</strong> — A Netflix-style video library for classic skate videos. HLS streaming, adaptive bitrate, collections, reactions. We’re uploading the films that defined skating — Yeah Right!, Sorry, This Is Skateboarding — and making them accessible in one place.</p><p><strong>Spots Map</strong> — Pin your local spots. Share them with your crew or keep them secret. Think Google Maps, but for ledges and rails.</p><p><strong>Trickipedia</strong> — An encyclopedia of every trick, organized by category and difficulty. See the progression. Track what you’ve landed. Know what’s next.</p><p>All of it. <strong>18,500 lines of code.</strong> Three codebases. One developer.</p><h3>The Tech Stack (For the Nerds)</h3><ul><li><strong>Backend:</strong> Node.js + Express, MongoDB Atlas, 25 API route files</li><li><strong>Web App:</strong> Next.js + React, deployed on AWS Amplify</li><li><strong>Mobile App:</strong> React Native (Expo), live on the iOS App Store</li><li><strong>Video Streaming:</strong> Bunny.net CDN with HLS adaptive streaming</li><li><strong>Real-time:</strong> Socket.io for DMs and live comments</li><li><strong>Payments:</strong> Stripe with freemium tiers</li><li><strong>Auth:</strong> JWT + Google SSO</li><li><strong>Maps:</strong> Google Places API</li></ul><p>The backend alone is 7,200+ lines. The web frontend is 9,100+ lines. The mobile app is 2,100+ lines.</p><p>A year ago, this would have taken a team. Today, it took me and Claude.</p><h3>How Coding Agents Changed Everything</h3><p>Here’s the workflow:</p><ol><li>I describe what I want in plain English</li><li>The agent writes the code</li><li>I review, test, iterate</li><li>Ship</li></ol><p>That’s it. That’s the whole process.</p><p>When I needed real-time messaging, I didn’t spend three days reading Socket.io docs. I told the agent: <em>“Add WebSocket support for direct messages. Emit events on new messages and read receipts. Store conversations in MongoDB.”</em></p><p>Fifteen minutes later, it worked.</p><p>When I needed video streaming, I didn’t research CDN providers for a week. I told the agent: <em>“Integrate Bunny.net Stream. Upload videos, generate HLS URLs, create signed tokens for playback.”</em></p><p>Done by lunch.</p><p>I’m not saying it’s <em>easy</em>. You still need to understand architecture. You still need to make product decisions. You still need to know when the agent is wrong (and it will be wrong).</p><p>But the grunt work? The boilerplate? The “how do I parse this response” and “what’s the right middleware pattern”?</p><p><strong>Gone. Automated. Handled.</strong></p><h3>Why Action Sports?</h3><p>Because we’ve been posting clips on platforms that don’t care about us.</p><p>Instagram’s algorithm buries your content unless you pay. TikTok is a slot machine designed to hijack your dopamine. YouTube shorts is… fine, I guess.</p><p>But none of them are <em>for</em> us. They’re for advertisers. We’re the product.</p><p>TrickBook is different. <strong>By riders, for riders.</strong></p><p>No ads. No algorithmic manipulation. No corporate shareholders demanding “engagement metrics.”</p><p>Just a community of people who love riding, building tools for each other.</p><h3>The Real Point</h3><p>This isn’t really about TrickBook. It’s about what’s now possible.</p><p>If I can build Instagram + WhatsApp + Netflix for action sports, what can <em>you</em> build?</p><ul><li>A Strava for rock climbers?</li><li>A Letterboxd for anime?</li><li>A Goodreads that doesn’t suck?</li></ul><p>The gatekeepers are gone. The tools are democratized. The only question is: <strong>what do you care enough to build?</strong></p><h3>Want to Build Like This?</h3><p>I’m documenting my entire process — the prompts, the architecture decisions, the mistakes, the wins.</p><h3>Get the Full Setup</h3><p>Sign up for my newsletter to get:</p><ul><li>🛠️ The exact coding agent setup I use</li><li>📋 Prompt templates that actually work</li><li>,️ Architecture patterns for solo developers</li><li>🚀 Weekly updates on what I’m shipping</li></ul><p>No spam. No fluff. Just the real stuff that’s working.</p><p>Because if one person can build a social platform, imagine what the rest of us can do.</p><p><em>Wes Huber is a full-stack developer and the creator of </em><a href="https://thetrickbook.com"><em>The TrickBook</em></a><em>. He’s been skating since he was 14 and coding since he was 19. He still can’t land a kickflip consistently.</em> <strong>#ByRidersForRiders</strong></p><p><em>Originally published at </em><a href="https://weshuber.com/blog/building-instagram-for-skaters.html"><em>https://weshuber.com</em></a><em> on February 9, 2026.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=98ea9d2838e8" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Building a Multi-Agent RAG System with LangGraph]]></title>
            <link>https://wesleybaxterhuber.medium.com/building-a-multi-agent-rag-system-with-langgraph-43071904b123?source=rss-cb71643b03ea------2</link>
            <guid isPermaLink="false">https://medium.com/p/43071904b123</guid>
            <category><![CDATA[multi-agent-systems]]></category>
            <category><![CDATA[ai-agent]]></category>
            <category><![CDATA[langchain]]></category>
            <dc:creator><![CDATA[Wesley Huber]]></dc:creator>
            <pubDate>Sat, 31 Jan 2026 19:21:37 GMT</pubDate>
            <atom:updated>2026-01-31T19:21:37.509Z</atom:updated>
            <content:encoded><![CDATA[<p>Why single-agent RAG falls apart — and how to fix it with specialized agents</p><p>I recently built a <strong>knowledge companion</strong> that needed to answer questions across multiple domains: health, finance, relationships, productivity, and more. A <em>naive RAG implementation failed spectacularly </em>— retrieving irrelevant chunks, confusing topics, and hallucinating answers when context was thin.</p><p><em>The fix</em>: a <strong>multi-agent architecture</strong> powered by LangGraph. Huge shoutout to LangChain Academy for making this accessible — their courses on agent orchestration were instrumental in getting this right.</p><p>Standing on the Shoulders of Giants: ReACT</p><p>Before diving into the architecture, it’s worth mentioning the ReACT (Reasoning + Acting) framework that underpins modern agentic LLMs. ReACT interleaves thinking and doing:</p><p>Thought → Action → Observation → Thought → Action…</p><p>Instead of blindly calling tools or purely reasoning in a vacuum, ReACT agents explicitly reason about what to do, take an action, observe the result, and adapt. This pattern is foundational to what LangGraph builds on.</p><p><strong>The Problem with Simple RAG</strong></p><p><strong>Traditional RAG</strong> follows a straightforward pattern: embed the question, retrieve top-k similar chunks, stuff them into context, generate response. This works fine for narrow domains. But when your knowledge base spans multiple topics with distinct retrieval needs, you run into issues:</p><p><em>Cross-contamination</em> — Health questions pulling finance docs, and vice versa</p><p><em>Diluted context</em> — Retrieving from everywhere means less relevant chunks per topic</p><p><em>No domain expertise</em> — A single retriever can’t optimize for the nuances of each knowledge area</p><p><em>Hallucination risk</em> — When context is thin or irrelevant, the LLM fills gaps with nonsense</p><p>I hit all four. My “helpful companion” confidently told a user to “diversify their emotional portfolio” when asked about managing anxiety. Not great.</p><p><strong>The Multi-Agent Solution</strong></p><p>Instead of one retriever to rule them all, what if each domain had its own specialist? That’s the core insight.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EBtN0V2pZ7YVNZuOoFibvQ@2x.jpeg" /></figure><p>The architecture has three key components:</p><ol><li><strong>The Supervisor Agent (Router)</strong></li></ol><p>This is the traffic controller. When a query comes in, the supervisor analyzes the question’s intent and domain, routes to the appropriate specialized agent, and handles multi-domain queries by coordinating multiple agents.</p><p><strong>2. Specialized RAG Agents</strong></p><p>Each domain gets its own agent with:</p><p>Dedicated vector store — Embeddings optimized for that domain</p><p>Domain-specific prompts — Context that helps the LLM understand the nuances</p><p>Tailored retrieval — Different chunk sizes, overlap, and k values per domain</p><p><strong>3. The Response Aggregator</strong></p><p>For queries that span multiple domains, the aggregator collects responses from relevant agents, synthesizes a coherent final answer, and cites which domain each piece of information came from.</p><p><strong><em>Why LangGraph?</em></strong></p><p>LangGraph makes this architecture surprisingly clean. Instead of spaghetti code with callbacks everywhere, you define:</p><p><strong>Nodes</strong> — Your agents and processing functions</p><p><strong>Edges</strong> — How control flows between them (including conditional routing)</p><p><strong>State</strong> — What gets passed around</p><p>The graph is explicit, debuggable, and easy to extend. Need a new domain? Add a node and an edge. Need to change routing logic? Update one function.</p><p><strong>Results</strong></p><p>After switching to multi-agent RAG, the improvements were dramatic:</p><p><em>Relevance score: 0.67 → 0.89</em></p><p><em>Hallucination rate: 23% → 4%</em></p><p><em>User satisfaction: 3.2/5 → 4.6/5</em></p><p>The improvement in relevance alone was worth the refactor. But the real win? No more “diversify your emotional portfolio” moments.</p><p><strong>When to Use This Pattern</strong></p><p><em>Multi-agent RAG shines when:</em></p><p>✅ Your knowledge base covers multiple distinct domains</p><p>✅ Different domains need different retrieval strategies</p><p>✅ You want specialized expertise per topic</p><p>✅ Single-agent RAG is mixing up unrelated content</p><p>Skip it when:</p><p>❌ You have a single, focused domain</p><p>❌ Your knowledge base is small enough for simple RAG</p><p>❌ Latency is critical (multi-agent adds round trips)</p><p>Getting Started</p><p>If you want to build something similar:</p><p>Take LangChain Academy’s courses — Especially the agent orchestration track</p><p>Start with the supervisor pattern — It’s the simplest multi-agent setup</p><p>Instrument everything — Use LangSmith to trace what’s happening</p><p>Iterate on routing — The supervisor’s classification is the critical path</p><p>The code examples for this implementation are available in the full tutorial. For production, you’ll want error handling, fallbacks, and proper state management. But the core pattern scales remarkably well.</p><p>Building something similar? I’d love to hear about it. Drop a comment below!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=43071904b123" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Deploying React 19 + Next.js to AWS Amplify with Environment Variables]]></title>
            <link>https://wesleybaxterhuber.medium.com/deploying-react-19-next-js-to-aws-amplify-with-environment-variables-a452a9637c82?source=rss-cb71643b03ea------2</link>
            <guid isPermaLink="false">https://medium.com/p/a452a9637c82</guid>
            <category><![CDATA[deployment]]></category>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[nextjs]]></category>
            <category><![CDATA[typescript]]></category>
            <dc:creator><![CDATA[Wesley Huber]]></dc:creator>
            <pubDate>Sun, 19 Oct 2025 15:39:41 GMT</pubDate>
            <atom:updated>2025-10-19T15:39:41.692Z</atom:updated>
            <content:encoded><![CDATA[<p><em>A “because it should’ve been easy, but it’s AWS” kind of guide.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vkTsOLFJ_2iC-IQowzgnNQ.png" /></figure><p><strong>Overview</strong><br> If you’ve ever tried deploying a shiny new Next.js app to AWS Amplify and thought, <em>“Huh, why are all my env vars mysteriously missing?”</em> — welcome to the club. Amplify is great for getting your site up fast, but when it comes to environment variables and Lambda runtimes, things get… let’s say… character-building.</p><p>This is a step-by-step guide to help you avoid yelling at your terminal at 2 AM.</p><p><strong>Prerequisites</strong></p><ul><li>Next.js 15+ with React 19</li><li>AWS Account (and infinite patience)</li><li>GitHub repo</li><li>Rudimentary knowledge of env vars and why AWS loves to ignore half of them</li></ul><h3>🪄 Step 1: Prepare Your App</h3><p><em>Because nothing screams “senior dev” like running </em><em>npm install for the 900th time.</em></p><ol><li><strong>Install Dependencies:</strong></li></ol><p>npm install next@latest react@latest react-dom@latest npm install @aws-sdk/client-dynamodb @aws-sdk/lib-dynamodb</p><p>You know the drill — node_modules gets bigger, your repo gets slower, and nothing is actually broken (yet).</p><h3>🫣 Step 2: Env Vars — The Plot Thickens</h3><p><em>Where dreams of </em><em>process.env working out-of-the-box go to die.</em></p><ol><li><strong>Create </strong><strong>.env.local.example:</strong></li></ol><pre># Database Configuration <br>DYNAMO_ACCESS_KEY_ID=your_access_key_here <br>DYNAMO_SECRET_ACCESS_KEY=your_secret_key_here <br>REGION=us-east-1  <br># Email Service <br>GMAIL_USER=your_email@domain.com <br>GMAIL_APP_PASSWORD=your_app_password  <br># Third-party APIs <br>WEB3FORMS_ACCESS_KEY=your_web3forms_key</pre><ol><li><strong>Update </strong><strong>next.config.ts:</strong></li></ol><pre>import type { NextConfig } from &#39;next&#39;<br>const nextConfig: NextConfig = {<br>  reactStrictMode: true,<br>  images: {<br>    domains: [],<br>    formats: [&#39;image/avif&#39;, &#39;image/webp&#39;],<br>  },<br>  env: {<br>    DYNAMO_ACCESS_KEY_ID: process.env.DYNAMO_ACCESS_KEY_ID,<br>    DYNAMO_SECRET_ACCESS_KEY: process.env.DYNAMO_SECRET_ACCESS_KEY,<br>    REGION: process.env.REGION,<br>    GMAIL_USER: process.env.GMAIL_USER,<br>    GMAIL_APP_PASSWORD: process.env.GMAIL_APP_PASSWORD,<br>    WEB3FORMS_ACCESS_KEY: process.env.WEB3FORMS_ACCESS_KEY,<br>  },<br>}<br>export default nextConfig</pre><p><strong>Why?</strong> Because if you don’t explicitly define these, Amplify’s Lambda runtime will pretend they never existed.</p><h3>🧾 Step 3: API Routes Meet AWS</h3><p><em>DynamoDB: Because we like typing a lot for a key-value store.</em></p><pre>// app/api/leads/route.ts<br>import { NextRequest, NextResponse } from &quot;next/server&quot;;<br>import { DynamoDBClient } from &quot;@aws-sdk/client-dynamodb&quot;;<br>import { DynamoDBDocumentClient, PutCommand } from &quot;@aws-sdk/lib-dynamodb&quot;;</pre><pre>// Initialize DynamoDB client with Amplify-compatible credentials<br>const accessKeyId = process.env.DYNAMO_ACCESS_KEY_ID || process.env.AWS_ACCESS_KEY_ID;<br>const secretAccessKey = process.env.DYNAMO_SECRET_ACCESS_KEY || process.env.AWS_SECRET_ACCESS_KEY;<br>const client = new DynamoDBClient({<br>  region: process.env.REGION || process.env.AWS_REGION || &quot;us-east-1&quot;,<br>  ...(accessKeyId &amp;&amp; secretAccessKey &amp;&amp; {<br>    credentials: {<br>      accessKeyId,<br>      secretAccessKey,<br>    },<br>  }),<br>});<br>const docClient = DynamoDBDocumentClient.from(client);<br>export async function POST(request: NextRequest) {<br>  try {<br>    const body = await request.json();<br>    const { email } = body;<br>    const command = new PutCommand({<br>      TableName: &quot;your-table-name&quot;,<br>      Item: {<br>        email: email.toLowerCase(),<br>        timestamp: new Date().toISOString(),<br>      },<br>    });<br>    await docClient.send(command);<br>    return NextResponse.json({ success: true });<br>  } catch (error) {<br>    console.error(&quot;Error:&quot;, error);<br>    return NextResponse.json({ error: &quot;Failed to process request&quot; }, { status: 500 });<br>  }<br>}</pre><p>This is where we <em>finally</em> use those env vars we fought so hard to configure.</p><h3>🧰 Step 4: Amplify Build Config</h3><p><em>Or how to spend an afternoon discovering that AWS won’t let you use </em><em>AWS_ prefixes.</em></p><p>Create amplify.yml:</p><pre>version: 1<br>frontend:<br>  phases:<br>    preBuild:<br>      commands:<br>        - npm ci --cache .npm --prefer-offline<br>    build:<br>      commands:<br>        # Debug: Show available environment variables<br>        - env | grep -E &#39;^(NEXT_PUBLIC_|GMAIL_|WEB3FORMS_|DYNAMO_|REGION)&#39; || true<br>        - npm run build<br>  artifacts:<br>    baseDirectory: .next<br>    files:<br>      - &#39;**/*&#39;<br>  cache:<br>    paths:<br>      - .next/cache/**/*<br>      - .npm/**/*</pre><p>And don’t forget to rename every AWS_ prefixed variable like a responsible adult.</p><h3>🚀 Step 5: Deploy</h3><p><em>Click button. Watch build fail. Realize you forgot to rename one env var. Cry. Repeat.</em></p><ol><li>Go to AWS Amplify Console.</li><li>“New app” → “Host web app”</li><li>Connect your GitHub repo</li><li>Select branch (usually main)</li><li>Add environment variables in <strong>App Settings → Environment Variables</strong>:</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*MR6RvUoIsNMhWgKzo2408A.png" /></figure><p><em>If it builds first try, congratulations — you’ve broken the laws of cloud physics.</em></p><h3>🐛 Step 6: Common “Features” (a.k.a. Problems)</h3><p><em>Welcome to the circus.</em></p><ul><li><strong>Env vars missing?</strong> AWS just didn’t feel like showing up today.</li><li><strong>Credential errors?</strong> Obviously your fault.</li><li><strong>TypeScript errors?</strong> Congrats, you’ve joined the Next.js 15 therapy group.</li></ul><p><strong>Solutions:</strong></p><ol><li>Ensure variables are declared in next.config.ts.</li><li>Never use AWS_ prefixes in Amplify env vars.</li><li>Add proper fallbacks in your SDK clients.</li><li>Read the build logs. No, really — read them.</li></ol><h3>🕵️ Step 7: Debug Like a Pro</h3><p><em>console.log — the real DevOps tool of the 21st century.</em></p><p>Add some noisy logs:</p><pre>export async function POST(request: NextRequest) {<br>  console.log(&quot;DYNAMO_ACCESS_KEY_ID exists:&quot;, !!process.env.DYNAMO_ACCESS_KEY_ID);<br>  console.log(&quot;Region:&quot;, process.env.REGION);<br>  // API logic goes here<br>}</pre><p>Then check AWS CloudWatch:</p><pre>aws logs tail /aws/amplify/your-app-id --region us-east-1 --since 10m</pre><h3>Best Practices</h3><ul><li>Don’t commit secrets. (Yes, even if you “just wanted to test it real quick.”)</li><li>Validate env vars early, or suffer later.</li><li>Logs are your friends. Until they’re not.</li></ul><h3>Conclusion</h3><p>Deploying React 19 + Next.js 15 on AWS Amplify is entirely possible… if you hold your tongue just right, never use AWS_ prefixes, and whisper sweet nothings to next.config.ts.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a452a9637c82" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[AI Agent Frameworks — Simplified]]></title>
            <link>https://wesleybaxterhuber.medium.com/ai-agent-frameworks-simplified-e49d22344734?source=rss-cb71643b03ea------2</link>
            <guid isPermaLink="false">https://medium.com/p/e49d22344734</guid>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[ai-agent-development]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Wesley Huber]]></dc:creator>
            <pubDate>Thu, 09 Oct 2025 09:48:24 GMT</pubDate>
            <atom:updated>2025-10-09T09:48:24.195Z</atom:updated>
            <content:encoded><![CDATA[<h3>AI Agent Frameworks — Simplified</h3><p>I’m here to give you the story on <strong>AI Agent Frameworks</strong>. In this article, we’ll break down the rise of the most disruptive early framework that shook up Web3, explore how big players have adopted the model for the enterprise, and look at newer, more playful frameworks that can inspire developers to <strong>build their own</strong>.</p><p>Classically developers build apps, and the users do the clicking, typing, and navigating.</p><p><strong>But a shift is happening.</strong></p><p>AI agents are beginning to <strong><em>reason</em></strong>, <strong><em>act</em></strong>, and <strong><em>collaborate</em></strong> — without us micromanaging every step.</p><blockquote>“Apps wait for users. Agents act.”</blockquote><p>And just like the web once had its Rails, React, and Django moment…<br><strong>AI agents are now having their framework moment.</strong></p><h3>What Is an AI Agent Framework?</h3><p>An <strong>AI Agent Framework</strong> is a blueprint for building AI Agent software that <em>thinks and acts autonomously</em>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0rKa3hDwlyBb_TScLZLTRA.jpeg" /></figure><p>Modern <strong>AI agent frameworks</strong> are built around a common conceptual model of how an <strong>autonomous agent</strong> operates.<br> The image above (from the <a href="https://arxiv.org/pdf/2308.11432">paper <em>“A Survey on LLM-based Autonomous Agents”</em></a>) lays out the fundamental components:</p><ul><li>🪪 <strong>Profile</strong> — defines who the agent is (personality, demographics, goals)</li><li>🧠 <strong>Memory</strong> — gives it continuity and context over time</li><li>🧭 <strong>Planning</strong> — allows it to reason, branch, and chain actions</li><li>🛠 <strong>Action</strong> — lets it act on the world through tools, APIs, or environments</li></ul><p>What <strong>agent frameworks</strong> do is <strong>operationalize this model at scale</strong>:</p><ul><li>Instead of a single agent, they allow <strong>multi-agent architectures</strong> where different agents can specialize, coordinate, and delegate.</li><li>Memory isn’t limited to one store — frameworks make it easy to plug in <strong>multiple memory sources</strong> (vector DBs, documents, on-chain data, telemetry streams).</li><li>Actions aren’t hardcoded — they expose <strong>tool interfaces</strong> and standardized protocols (e.g., OpenAPI, MCP) so new capabilities can be added modularly.</li><li>Planning can be <strong>distributed</strong> — one agent can handle reasoning, another can execute, another can monitor outcomes.</li></ul><blockquote><strong><em>In short:</em></strong><em> agentic frameworks take the classic Profile–Memory–Planning–Action architecture and make it </em><strong><em>composable, extensible, and multi-agent</em></strong><em>.</em></blockquote><p>This is the foundation that powers systems like <strong>ElizaOS</strong>, <strong>Microsoft Agent Framework</strong>, and <strong>PanduAI</strong> — each with its own flavor, but built on these same pillars.</p><ul><li>📝 Define agent behavior, personality, and reasoning</li><li>🔌 Plug into APIs, blockchains, or internal systems</li><li>🧠 Give agents memory and context</li><li>🔁 Orchestrate multi-agent workflows</li><li>📦 Deploy and scale easily</li></ul><p><strong>Typical stack layers:</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uem40yqrtBOQJxwv44jmng.png" /></figure><blockquote>If React helped us build better UIs, AI agent frameworks will help us build better autonomous systems.</blockquote><h4>ElizaOS: How Web3’s First Major Agent Framework Broke Out</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jYRyrr4OVmhQxiTQfL59qQ.png" /></figure><p><a href="https://elizaos.ai/"><strong>ElizaOS</strong></a> was one of the first agent frameworks to <em>disrupt an industry</em>, in this case, crypto. By creating a framework to develop autonomous agents that could easily plug into communities like Discord, Telegram and social media platforms like X (twitter), we saw an explosion of AI Agent memecoins — and we saw the memecoin behid the ElizaOS project skyrocket to a <strong><em>$2 Billion Dollar market cap</em></strong> in January of this year.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Qx2sa-5OFd2zxMFb76dA2A.png" /></figure><p>Another measure of success is the fact that ElizaOS was trending on GitHub and currently as about 17,000 stars on the repository which is insane.</p><h3>Key capabilities:</h3><ul><li>Web3-native plugin system (EVM chains, signing, smart contract calls) — Agents can trade on chain with their own wallet.</li><li>Multi-agent orchestration with persistent memory</li><li>Discord, Telegram, X integrations and more</li><li>Personality-driven behavior through JSON configuration</li><li>Confidential execution via Intel TDX enclaves</li></ul><p>For the first time, communities could launch <strong>autonomous DAO delegates</strong>, <strong>memecoin mascots</strong>, and <strong>trading bots</strong> with personality.</p><blockquote>Eliza didn’t just chat. Eliza participated.</blockquote><p>This early disruption came with risks — memory poisoning, malicious context injection, and the need for stronger security models. Still, ElizaOS remains a blueprint for how open-source frameworks can change an entire industry overnight.</p><h4>Microsoft Agent Framework: Enterprise Embraces the Model</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3xmAwEX2YsnGjjJlv_yt5Q.jpeg" /></figure><p>Where Eliza emerged from the grassroots of crypto culture, Microsoft has <a href="https://azure.microsoft.com/en-us/blog/introducing-microsoft-agent-framework/">brought the agent framework revolution</a> to the <strong>corporate world</strong>.</p><p>The <strong>Microsoft Agent Framework</strong> unifies their previous work on Semantic Kernel and AutoGen into a single, open-source runtime designed for <strong>enterprise, production-grade systems</strong>.</p><p>Microsoft is already citing enterprise pilots (<a href="https://devblogs.microsoft.com/foundry/introducing-microsoft-agent-framework-the-open-source-engine-for-agentic-ai-apps/">BMW, KPMG, Commerzbank</a>) using Agent Framework + Azure Foundry.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*o7bKthjacKw83qysmB7Zdg.png" /></figure><h3>What it offers:</h3><ul><li>Graph-based <strong>multi-agent orchestration</strong></li><li>Pluggable memory and context providers</li><li>Standardized tool calling via <strong>MCP</strong> and <strong>OpenAPI</strong></li><li>Built-in observability, telemetry, and debugging tools</li><li>Governance, security, and compliance baked in</li></ul><p>This framework is great for integrating into Microsoft products like Teams, and working with an enterprise stack built on Azure.</p><blockquote>Eliza moved fast and broke things. Microsoft wants to make sure you can build things that won’t break.</blockquote><p>It’s not Web3-native, but it’s an enterprise Microsoft-ready powerhouse for structured agent systems at scale.</p><h4>PanduAI: The Playful Frontier of Voice, Chat &amp; Art</h4><p>Not every agent framework needs to be enterprise or DeFi-grade.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BIJPrvW6CrdZCvmSPu7YQA.jpeg" /></figure><p>Enter <strong>PanduAI</strong>, a <a href="https://github.com/pandupandas/panduAi-core">modular AI framework</a> behind the <a href="https://pandupandas.com/">PanduPandas</a> memecoin / NFT ecosystem.</p><p>PanduAI blends <strong>conversation</strong>, <strong>voice</strong>, and <strong>generative art</strong> into a lightweight, hackable stack.</p><h3>Core Modules:</h3><ul><li>🧠 <strong>PanduCore</strong> → conversational engine with memory &amp; personality</li><li>🗣 <strong>PanduVoice</strong> → ultra-low latency speech synthesis and recognition</li><li>🎨 <strong>PanduCanvas</strong> → generative art &amp; avatar pipeline</li><li>🧩 <strong>PanduOS</strong> → orchestrator connecting all modules</li></ul><p>PanduAI is not trying to build the next enterprise workflow platform. It’s creating <strong>playful, cultural agents</strong> that live in communities — memecoins, NFTs, social spaces.</p><blockquote>Pandu isn’t trying to be Skynet. Pandu is trying to make the internet more alive</blockquote><h3><strong>Comparing the Frameworks</strong></h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*m2lwmiVMyTMpmjP11Brakw.png" /></figure><h3>Why Developers Should Care</h3><p>AI agents aren’t just a curiosity. They’re a <strong>new computing layer</strong>.</p><p>Think about where agents can disrupt:</p><ul><li>🏭 <strong>Manufacturing</strong> — autonomous factory monitors and alerting agents</li><li>💰 <strong>Finance &amp; Crypto</strong> — DAO delegates, auto-trading bots, liquidity agents</li><li>🛍 <strong>E-commerce</strong> — product experts with long-term memory</li><li>🧑‍🎨 <strong>Media &amp; Art</strong> — voice + generative art personalities</li><li>🧑‍🏫 <strong>Education</strong> — adaptive tutors with reasoning</li><li>🧑‍⚕️ <strong>Healthcare</strong> — workflow copilots for practitioners</li></ul><p>Every major software wave had a framework moment. This is the <strong>agent framework moment</strong>.</p><h3>🛠 How to Build Your Own Agent Framework</h3><blockquote>“Every framework started as someone’s weekend project.”</blockquote><p>You don’t need a 100-person research team to build something powerful. You need curiosity, good abstractions, and a clear use case.</p><p>Here’s a simple progression:</p><ol><li><strong>Start with an agent core</strong> → reasoning + reply</li><li><strong>Add tools</strong> → API calls, chain actions, functions</li><li><strong>Add memory</strong> → short-term context + vector store</li><li><strong>Add orchestration</strong> → routing between agents</li><li><strong>Add interfaces</strong> → voice, chat, canvas</li><li><strong>Secure it</strong> → guardrails, logging, permissions</li><li><strong>Instrument it</strong> → telemetry, monitoring, testing</li></ol><p>Your agent doesn’t need to be the next Eliza or Pandu.<br>It just needs to <em>work well for your problem space</em>.</p><h3>The Future Is Agentic</h3><ul><li>ElizaOS showed how open-source frameworks can shake up entire industries.</li><li>Microsoft is industrializing the agent stack for enterprise.</li><li>PanduAI proves agents can also be cultural, fun, and weirdly lovable.</li></ul><p>Frameworks are just the beginning.</p><p>The real wave will come from <strong>developers</strong> who build new experiences, industries, and movements on top of these foundations.</p><blockquote>“Apps were built by startups. Agents will be built by everyone.”</blockquote><p>Thanks for reading, if you gained some value from this article please consider <a href="https://buymeacoffee.com/wesleyhuber">Buying me a Coffee</a> — cheers!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e49d22344734" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Deploy a Serverless API with AWS SAM & GitHub Actions — Step-by-Step Guide]]></title>
            <link>https://wesleybaxterhuber.medium.com/deploy-a-serverless-api-with-aws-sam-github-actions-step-by-step-guide-42a72e9613bd?source=rss-cb71643b03ea------2</link>
            <guid isPermaLink="false">https://medium.com/p/42a72e9613bd</guid>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[serverless]]></category>
            <category><![CDATA[github]]></category>
            <category><![CDATA[api]]></category>
            <dc:creator><![CDATA[Wesley Huber]]></dc:creator>
            <pubDate>Fri, 28 Feb 2025 14:07:44 GMT</pubDate>
            <atom:updated>2025-02-28T14:18:25.776Z</atom:updated>
            <content:encoded><![CDATA[<h3>Deploy a Serverless API with AWS SAM &amp; GitHub Actions — Step-by-Step Guide</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mcL74FIdttZlj76GonlQEA.png" /><figcaption>Deploy a Serverless API with AWS SAM &amp; GitHub Actions — Step-by-Step Guide</figcaption></figure><h3>Table of Contents</h3><p>– Intro: Why use SAM to deploy an API?<br>– Prerequisites: Install SAM CLI<br>– Step 1: Initialize SAM <br>– Step 2: Define the API Endpoint<br>– Step 3: Build and Test Locally<br>– Step 4: Deploy the API<br>– Step 5: CI / CD With Github Actions<br>– Conclusion</p><h3>Intro: Why use SAM to deploy an API?</h3><p>Deploying an API using AWS Serverless Application Model (SAM) is an efficient and scalable approach for cloud-based applications.</p><p>It simplifies infrastructure management, provides built-in SSL/TLS support, and seamlessly integrates with AWS services such as CloudFormation, Lambda, S3, Route 53, and CloudWatch.</p><p>By leveraging a serverless model, developers can focus on writing application logic while AWS handles scaling, security, and maintenance, making it an ideal solution for modern, high-availability APIs.</p><p>This tutorial will guide you step-by-step on how to create a simple API, deploy it using AWS CloudFormation with the SAM CLI, and then automate deployments using GitHub Actions.</p><p>Here’s a more detailed diagram of this workflow:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/808/1*xVOKmP03HnIU4O4VJ4V-1w.png" /></figure><p><strong>Prerequisites</strong></p><ul><li>AWS CLI and SAM CLI installed.</li><li><a href="https://medium.com/@wesleybaxterhuber/how-to-install-node-js-on-macos-and-windows-in-2025-with-nvm-55c54692869a">Node.js installed</a></li><li><a href="https://www.docker.com/">Docker</a> for local testing.</li><li><a href="https://github.com/">GitHub</a> account.</li></ul><h3>Prerequisites: Install the SAM CLI</h3><p>Before diving into the tutorial, ensure the AWS SAM CLI is installed on your machine. Below are the installation steps for macOS and Windows:</p><h4>For macOS</h4><p><strong>Install Homebrew</strong> (if not already installed):</p><pre>/bin/bash -c “$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/HEAD/install.sh)&quot;</pre><p><strong>Add the AWS tap and install the SAM CLI</strong>:</p><pre>brew tap aws/tap<br>brew install aws-sam-cli</pre><p>Verify the installation with sam --version</p><h4>For Windows</h4><p><strong>Download the SAM CLI installer</strong>:</p><ul><li>Go to the <a href="https://docs.aws.amazon.com/serverless-application-model/latest/developerguide/install-sam-cli.html">official AWS SAM CLI page</a> and download the latest Windows installer.</li></ul><p><strong>Run the installer</strong>:</p><ul><li>Double-click the downloaded .exe file and follow the installation wizard.</li></ul><p><strong>Add SAM CLI to the PATH</strong> (if not automatically added):</p><ul><li>Open the Start menu and search for “Environment Variables.”</li><li>Click on “Edit the system environment variables.”</li><li>Under “System Properties,” click the “Environment Variables” button.</li><li>In “System Variables,” find Path and click &quot;Edit.&quot;</li><li>Add the directory where the SAM CLI was installed (e.g., C:\Program Files\Amazon\AWS SAM CLI\bin).</li></ul><p><strong>Verify the installation</strong>: Open a command prompt and run sam --version</p><h3>Step 1: Initialize the SAM CLI</h3><p><strong>Create a new repository on your machine:</strong> mkdir aws-serverless-api</p><p><strong>Navigate to the directory: </strong>cd aws-serverless-api</p><p><strong>Initialize a new SAM application: </strong>sam init <br>then you’ll be prompted with a few options, here is my setup:</p><ul><li>Choose <strong><em>“1 — AWS Quick Start Templates.”</em></strong></li><li>Choose <strong><em>“7 — Serverless API”</em></strong><em> </em>to deploy a Lambda backed serverless API on API Gateway”</li><li>Choose <strong><em>“Node.js 20.x”</em></strong></li><li>Choose <strong><em>“N”</em></strong> or No for X-Ray tracing since we want to reduce our costs and this is just a simple Hello World example. If you are making a complex API pulling in different services you might want to enable that to help in debugging.</li><li>I’m also choosing <strong><em>“N”</em></strong> for cloudwatch monitoring since we don’t need analytics on performance for this Hello World example</li><li>Also <strong><em>“N”</em></strong><em> </em>for the json format in the lambda log since this is just a Hello World example</li></ul><p>Now let’s cd into the directory that the sam init generated, I chose to name my project sam-hello-world so let’s go:</p><p>cd sam-hello-world</p><p>Then open the project in VSCode code . — when you open the project in VSCode or your favorite IDE, you should already see a lot of generated files.</p><h4>Understanding the Generated Files</h4><p>Here&#39;s what each of the files generated from sam init does:</p><ul><li><strong>template.yaml</strong> – Defines your API Gateway, Lambda, and other AWS resources.</li><li><strong>src/handlers/</strong> – Contains sample Lambda function handlers.</li><li><strong>events/</strong> – Sample test events for local testing.</li><li><strong>package.json</strong> – Manages dependencies for Node.js Lambda functions.</li><li><strong>.gitignore, </strong><strong>README.md</strong> – Standard project setup files.</li></ul><p>We will <strong>modify and remove some of these files to fit our very simple “hello world” API</strong>.</p><h3>Step 2: Define the API Endpoint</h3><p>Now, open template.yaml and replace the existing Lambda function definitions with the following:</p><pre>AWSTemplateFormatVersion: &quot;2010-09-09&quot;<br>Transform: AWS::Serverless-2016-10-31<br>Description: AWS SAM template for deploying Node.js/Express API to Lambda<br><br>Resources:<br>  HelloWorldFunction:<br>    Type: AWS::Serverless::Function<br>    Properties:<br>      Handler: src/app.lambdaHandler<br>      Runtime: nodejs20.x<br>      Events:<br>        HelloWorld:<br>          Type: Api<br>          Properties:<br>            Path: /<br>            Method: GET</pre><p>This defines one route at the root of our API that is a GET request</p><h4>Update the Lambda Function</h4><p>In the /src directory let’s create a file called app.js</p><p>in src/app.js let’s write a function to output hello world from our route:</p><pre>exports.lambdaHandler = async (event) =&gt; {<br>    return {<br>        statusCode: 200,<br>        body: JSON.stringify({ message: &quot;Hello, World!&quot; })<br>    };<br>};</pre><h4>Remove Unnecessary Files</h4><p>Now, let’s remove the additional handlers and events that the sam init generated in step 1.</p><p>Open a terminal in the project directory and <strong>execute these commands</strong>:</p><pre>rm -rf src/handlers/get-all-items.mjs src/handlers/get-by-id.mjs src/handlers/put-item.mjs<br>rm -rf events/</pre><h3>Step 3: Build and Test Locally</h3><p>Open a terminal in the project directory and follow the steps below:</p><ol><li><strong>Build the application</strong>:</li></ol><ul><li>sam build you should see an output like this:</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CFDpexQc1XcNSSiPyZ1ZTg.png" /></figure><ol><li><strong>Start a local API server</strong>:</li></ol><ul><li>sam local start-api — Note you’ll need to have <a href="https://www.docker.com/">docker</a> installed to run this locally. When successful you should see a container created for our API in docker desktop</li></ul><ol><li>Open a browser or use curl in the terminal to test the endpoint:</li></ol><ul><li>curl <a href="http://127.0.0.1:3000/">http://127.0.0.1:3000/</a></li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*dlcGwHLlvuh6-iLoSK614Q.png" /></figure><p>If you see the JSON returned from our endpoint then you’ve successfully started the API locally! 🚀</p><h3>Step 4: Deploy the API</h3><p><strong>Package and deploy the API using SAM</strong>:<br>Now that we’ve confirmed our API is working locally let’s deploy it to AWS using sam deploy --guided</p><p>This guided deployment will allow you to:<br>1. Choose your stack name, mine is sam-hello-world <br>2. Choose AWS Region, mine is us-east-1 <br>3. Confirm changes before deploy, I chose Y <br>4. Allow SAM-CLI Role Creation, Y <br>5. Disable Rollback, choose N we want the tool to clean up any failed resources<br>6. HelloWorldFunction has no Authentication, is this okay? For the purposes of this tutorial Y <br>7. Save arguments to configuration file Y <br>8. Configuration file samconfig.toml <br>9. SAM configuration environment — leave this as default</p><p>If your deployment was successful you should see the resources successfully created like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3b1Rkj4T-lKTUrYuQHATQA.png" /></figure><p>When the deployment is successful, <a href="https://console.aws.amazon.com/">login to your AWS account</a>, and search for the API Gateway service, go to API settings, copy your domain paste it in the URL input on your browser window and add `/Prod` to the end of your URL to reach your deployed API endpoint.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*QVVQlGaUDSI5LnQlPYlbLQ.png" /></figure><p>My URL looks like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Ne-a0mNkwFRAtYu5lSnuYw.png" /></figure><p>You can also search for Lambda in the AWS console to view your deployed serverless Lambda function where the code we wrote for this endpoint lives.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*o6GW6bkj7wqW6MvbqiYubg.png" /></figure><p>You should also see an S3 bucket created which holds the bundle of our code. Now that we’ve successfully deployed our API, let’s automate this process with GitHub so that anytime we want to develop on our API the deployment can automatically trigger by just commiting to the MAIN or MASTER branch.</p><h3>Step 5: CI / CD With Github Actions</h3><h4>Initialize your Git repo and push to your master branch</h4><p>First we’ll initialize a git repository in our project’s directory with git init <br>Then let’s add these files to our .gitignoreas we do not want to commit them.</p><pre>echo &quot;node_modules/\n.aws-sam/\nsamconfig.toml&quot; &gt;&gt; .gitignore</pre><p>Now we’re safe to stage all of the files in the project: git add . and then make our first commit git commit -m &quot;Initial commit - AWS SAM API setup&quot;</p><p>This next step assumes you have the GitHub CLI installed — This step is so we can create a GitHub repo from our current directory:</p><pre>gh repo create aws-sam-cli-tutorial --public --source=. --remote=origin</pre><p>If you don’t have the GitHub CLI you can create the repo manually on github and then point the directory to your repo with this command:</p><pre>git remote add origin https://github.com/YOUR_USERNAME/YOUR_REPO_NAME.git</pre><p>Finally, let’s push our code to GitHub git push -u origin master</p><h4>Set Up GitHub Actions for CI/CD</h4><p>Let’s create the GitHub actions workflow file:</p><pre>mkdir -p .github/workflows<br>touch .github/workflows/deploy.yml</pre><p>Open that deploy.yml file in vscode and paste the following yaml configuration:</p><pre>name: Deploy API to AWS CloudFormation<br><br>on:<br>  push:<br>    branches:<br>      - master<br><br>jobs:<br>  deploy:<br>    runs-on: ubuntu-latest<br><br>    steps:<br>      - name: Checkout Code<br>        uses: actions/checkout@v3<br><br>      - name: Install AWS SAM CLI<br>        uses: aws-actions/setup-sam@v2<br><br>      - name: Configure AWS CLI<br>        uses: aws-actions/configure-aws-credentials@v2<br>        with:<br>          aws-access-key-id: ${{ secrets.AWS_ACCESS_KEY_ID }}<br>          aws-secret-access-key: ${{ secrets.AWS_SECRET_ACCESS_KEY }}<br>          aws-region: us-east-1<br><br>      - name: Build and Deploy API<br>        run: |<br>          sam build<br>          sam deploy \<br>          --stack-name sam-hello-world \<br>          --s3-bucket ${{ secrets.AWS_S3_BUCKET }} \<br>          --capabilities CAPABILITY_IAM \<br>          --region ${{ secrets.AWS_REGION }} \<br>          --no-confirm-changeset \<br>          --no-fail-on-empty-changeset</pre><p>In this .yaml file for automated deployment, we rely on the AWS-Actions repositories from GitHub: <a href="https://github.com/orgs/aws-actions/repositories">https://github.com/orgs/aws-actions/repositories</a> to install the SAM CLI and configure our credentials — these AWS repos may change in a few years so update as needed.</p><p>Now let’s go to github.com — go to your repositories, click the repository we made for this project, go to settings, actions, and you’ll see this screen to add environment and repository secrets.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*b9RSHSnmY7YG1Tw_AlMEyA.png" /></figure><p>We’ll want to add Repository secrets for:<br><strong>AWS_ACCESS_KEY_ID<br>AWS_SECRET_ACCESS_KEY<br>AWS_REGION <br></strong><strong>AWS_S3_BUCKET</strong></p><p>To get these values, go to your AWS console. For <strong>AWS_REGION</strong> you’ll want to use the same region that you deployed from your local machine with. If you are unsure of the region, search for cloudformation and find your stack that you created, click on it and you should see the region in the URL.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mTNuLdRqVlY60nHSlmir5A.png" /></figure><p>For the <strong>AWS_S3_BUCKET </strong>search for S3 in the AWS console and select the bucket that was created when we deployed from our machine in Step 4. Use the ID of this bucket for your value, mine is shown in the screenshot below.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6E8Xm1T1__5fkGHf2FHcrw.png" /></figure><p>For the <strong>AWS_ACCESS_KEY_ID </strong>and <strong>AWS_SECRET_ACCESS_KEY </strong>go to the IAM service in the AWS console:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*gc4YOlmqc8mYOIEgdyw4Kg.png" /></figure><p>From here create an IAM user if you don’t already have one with access you want to use, and add at least these permission scopes to the user:<br>AWSLambdaFullAccess IAMFullAccess AmazonS3FullAccess CloudFormationFullAccess AmazonAPIGatewayAdministrator</p><p>You can also just add <strong>AdministratorAccess </strong>if you’re comfortable with that. After that we’ll want to create an Access key and Secret for this user. Choose the CLI as the use case for the access key. Make sure to copy your secret because you won’t be able to see it again.</p><p>Now, go back to GitHub and add your repository secrets:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-5KW7f4VZZh18euZAfrlSw.png" /></figure><p>Okay, final step is to test that everything works! Update your app.js to say something else, I changed mine to “What up, World!”</p><pre>exports.lambdaHandler = async (event) =&gt; {<br> return {<br>  statusCode: 200,<br>  body: JSON.stringify({ message: &quot;What up, World!&quot; }),<br> };<br>};</pre><p>Then- push your changes to your master branch. git add . git commit -m &#39;updated message&#39; git push — make sure your on the master branch when you push.</p><p>In your GitHub repo, you should see the action run and successfully deploy:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GrE8xledEv7-5cjalWI8Qw.png" /></figure><p>Finally go back to your web browser and paste the URL from the API Gateway we set up. Mine is: <a href="https://tt40c6vgm3.execute-api.us-east-1.amazonaws.com/Prod">https://tt40c6vgm3.execute-api.us-east-1.amazonaws.com/Prod</a> — and confirm the message was changed.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*A8eOO2QA7UGYqRilSJNLtw.png" /></figure><h3>Conclusion</h3><p>Congratulations! You just created a CI/CD deployment pipeline for a serverless AWS Stack, now you can get busy building out your API with useful services. Let me know if this article was useful, if you have any questions or suggestions for improvements!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=42a72e9613bd" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How to Install Node.js on macOS and Windows in 2025 with NVM]]></title>
            <link>https://wesleybaxterhuber.medium.com/how-to-install-node-js-on-macos-and-windows-in-2025-with-nvm-55c54692869a?source=rss-cb71643b03ea------2</link>
            <guid isPermaLink="false">https://medium.com/p/55c54692869a</guid>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[nodejs]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[web-app-development]]></category>
            <dc:creator><![CDATA[Wesley Huber]]></dc:creator>
            <pubDate>Mon, 27 Jan 2025 13:51:33 GMT</pubDate>
            <atom:updated>2025-01-27T13:51:33.174Z</atom:updated>
            <content:encoded><![CDATA[<p>Node.js is essential for modern web and backend development. Whether you’re new to development or just need a refresher, this guide will walk you through the installation process on macOS and Windows. NVM, or Node Version Manager, allows you to manage multiple node versions on your machine.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*mMMyAicg9rkUfEZFhvgBQA.png" /></figure><h3>Installing Node.js on macOS</h3><ol><li>Install nvm</li></ol><p>Open your terminal and run the following command to download and install nvm:</p><pre>curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.5/install.sh | bash</pre><p>After installation, update your shell configuration file to load nvm. Based on your shell:</p><p>For <strong>zsh</strong> (default on macOS Catalina and later):</p><pre>nano ~/.zshrc</pre><p>Add the following lines:</p><pre>export NVM_DIR=&quot;$HOME/.nvm&quot;<br>[ -s &quot;$NVM_DIR/nvm.sh&quot; ] &amp;&amp; \. &quot;$NVM_DIR/nvm.sh&quot; # This loads nvm</pre><p>For <strong>bash</strong> (older macOS versions):</p><pre>nano ~/.bash_profile</pre><ul><li>Add the same lines as above</li></ul><p>Save and reload your shell configuration:</p><pre>source ~/.zshrc  # Or source ~/.bash_profile for bash</pre><p>Verify that nvm was installed:</p><pre>nvm --version</pre><h3>2. Install Node.js Using nvm</h3><p>Install the latest LTS (Long Term Support) version of Node.js:</p><pre>nvm install --lts</pre><p>Use the installed version:</p><pre>nvm use --lts</pre><p>Verify your Node.js and npm versions:</p><pre>node -v<br>npm -v</pre><h3>Installing Node.js on Windows</h3><ul><li><strong>Download </strong><strong>nvm-windows</strong>:</li><li>Go to the official <a href="https://github.com/coreybutler/nvm-windows/releases">nvm-windows GitHub page</a>.</li><li>Download the latest .zip file (e.g., nvm-setup.zip).</li><li><strong>Install </strong><strong>nvm-windows</strong>:</li><li>Unzip and run the nvm-setup.exe file.</li><li>Follow the installation wizard and select the default options.</li><li><strong>Verify </strong><strong>nvm-windows Installation</strong>: Open the Command Prompt and run: nvm version If the installation was successful, you will see the installed nvm version.</li><li><strong>Install Node.js Using </strong><strong>nvm</strong>:</li></ul><p>Install the latest LTS (Long Term Support) version of Node.js:</p><pre>nvm install lts</pre><p>Use the installed version:</p><pre>nvm use lts</pre><p>Verify the installation:</p><pre>node -v<br>npm -v</pre><p><strong>Switch Between Node.js Versions</strong> (Optional): If you need to use a different version for a project, you can install it and switch:</p><pre>nvm install 16<br>nvm use 16</pre><h3>Why Use nvm?</h3><p>Using nvm ensures flexibility:</p><ul><li>You can easily switch between Node.js versions for different projects.</li><li>Avoid conflicts with system-wide Node.js installations.</li><li>Test your application across multiple Node.js versions.</li></ul><p>A lot of times certain projects, especially legacy projects, require a specific version of node.js to run properly. NVM is the best way to switch between versions on your local machine. I hope this tutorial helped you install node.js on your computer, let me know in the comments if you have any questions- I’ll do my best to respond~ Peace out and happy coding!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=55c54692869a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Creating a CI/CD Pipeline using Github and AWS EC2]]></title>
            <link>https://wesleybaxterhuber.medium.com/creating-a-ci-cd-pipeline-using-github-and-aws-ec2-f1f9a94bf4e2?source=rss-cb71643b03ea------2</link>
            <guid isPermaLink="false">https://medium.com/p/f1f9a94bf4e2</guid>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[aws]]></category>
            <category><![CDATA[github]]></category>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[automation]]></category>
            <dc:creator><![CDATA[Wesley Huber]]></dc:creator>
            <pubDate>Tue, 30 Apr 2024 12:39:23 GMT</pubDate>
            <atom:updated>2024-04-30T12:39:23.677Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*-MgUSQxIQyNYx32-_p9IJg.png" /></figure><p>Hey everyone, today I want to do a tutorial of one of my favorite ways to deploy a quick REST API, using NodeJS / Express, to an Amazon EC2 server. This technique essentially enables your EC2 server to respond to changes pushed to your Github repository.</p><h3>First things first, “Hello World”</h3><p>Create a new directory on your machine for a NodeJS / Express app to live. I’m calling mine rekognition</p><p>Open up a terminal and run this command to initialize a package.json npm init -y</p><p>Install express.js npm i express</p><p>Create a basic express server with a ‘hello world’ at the default route</p><pre>const express = require(&#39;express&#39;);<br>const app = express();<br>const port = 3000;<br><br>app.get(&#39;/&#39;, (req, res) =&gt; {<br>    res.send(&#39;Hello World!&#39;);<br>});<br><br>app.listen(port, () =&gt; {<br>    console.log(`Example app listening at http://localhost:${port}`);<br>});</pre><p>Test to make sure your are seeing ‘Hello World!’ at <a href="http://localhost:3000/">http://localhost:3000/</a></p><p>Good job, step one complete! 💪</p><h3>Next let’s push our code to Github</h3><p>Initialize a Git repo in your directory and make your first commit —</p><pre>git init<br>git add .<br>git commit -m &quot;First commit&quot;</pre><p>Create a remote repository on your Github account</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/754/1*QUGTkcSQf82881RhiOEyFA.png" /></figure><p>Add the new repo as the origin and push your branch</p><pre>git remote add origin &lt;REMOTE_REPOSITORY_URL&gt;<br>git push -u origin master</pre><p>Replace &lt;REMOTE_REPOSITORY_URL&gt; with the link to your remote repo that you just created.</p><p>You might need to change your repository settings to use ‘master’ as the default branch instead of ‘main’</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6DlrHrUCJDS2iyxdWAI-gA.png" /></figure><p>After you do this you can delete the ‘main’ branch.</p><p>Step 2: complete! We’re cooking with gas ⛽️</p><h3>Let’s go to the cloud — setup the Ec2 server</h3><p>Here are the steps to launch your EC2 Instance:</p><ul><li>Log in to your <a href="https://console.aws.amazon.com">AWS Management Console</a>.</li><li>Navigate to the EC2 dashboard and launch a new instance.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*NIxpypA-B8cDdRcDGp5syw.png" /></figure><ul><li>Choose an appropriate AMI (Amazon Machine Image), I am using Amazon Linux 2.</li><li>Select an instance type (e.g., t2.micro if you are under the free tier).</li><li>Configure instance details, add storage, and add tags as necessary.</li><li>Configure a security group to allow inbound traffic on ports you’ll need (at least TCP port 22 for SSH and your application’s port, and then 3000 for web traffic).</li><li>Review and launch the instance, ensure that you save the key pair used for SSH access.</li></ul><p>Now let’s test logging into your server via SSH on a terminal:</p><p>ssh ec2-user@&lt;YOUR_SERVER_IP&gt; -i &lt;YOUR_PRIVATE_KEY&gt;.pem</p><p>Replace the variables there with the values you defined for your EC2 instance.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8liOONdqsqJCYkECLwUdwA.png" /></figure><p>If you see this you’ve connected to your server, congrats! 👏</p><h3>We’re not done just yet though — we’re just getting started 😏</h3><p>Let’s install NodeJS on our server:<br>sudo yum install -y nodejs</p><p>Install <a href="https://pm2.keymetrics.io/">pm2</a> globally on our server via npm</p><p>sudo npm i -g pm2</p><p>Install git on the server and configure your credentials:</p><p>sudo yum install git -y</p><pre>git config --global user.name &quot;Your Name&quot;<br>git config --global user.email &quot;your.email@example.com&quot;</pre><p>Okay now we’re going to create a var/www/ folder if it does not already exist:</p><p>sudo mkdir -p /var/www</p><p>Then add ownership to our ec2-user:</p><p>sudo chown -R ec2-user:ec2-user /var/www</p><p>Navigate to the directory <em>from the root of your server</em> — not the /home folder people 👌 cd ~/var/www</p><p>Now from here we will clone our git repository: git clone <a href="https://github.com/USERNAME/REPO_NAME.git">https://github.com/USERNAME/REPO_NAME.git</a> YOUR_PROJECT_DIRECTORY</p><p>Replace [USERNAME] [REPO_NAME] and [YOUR_PROJECT_DIRECTORY] with your own values.</p><p>Next we change directory into our repository, so cd YOUR_PROJECT_DIRECTORYcand reate the ecosystem.config.js file for our pm2 setup:</p><p>nano ecosystem.config.js</p><p>It looks like this:</p><pre>module.exports = {<br>  apps : [{<br>    name: &quot;my-app&quot;,  // A name for your application<br>    script: &quot;app.js&quot;,  // The script file to launch the app<br>    instances: &quot;max&quot;,  // This can be set to the number of CPU cores to use, or &quot;max&quot; to use all available<br>    autorestart: true,  // Automatically restart app if it crashes<br>    watch: false,  // Enable/disable watching file changes for automatic restart<br>    max_memory_restart: &#39;1G&#39;,  // Optional: Restart app if it reaches a memory limit<br>    env: {<br>      NODE_ENV: &quot;development&quot;,<br>    },<br>    env_production: {<br>      NODE_ENV: &quot;production&quot;,<br>    }<br>  }]<br>};</pre><p>Coolbeans, okay — once that is created and saved, let’s start up the app with pm2:</p><p>pm2 start ecosystem.config.js</p><p>Now we can check that we have a ‘Hello World!’ from our server on port 3000:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*z1itmc25RVLQ0Hsenp4FJQ.png" /></figure><p>To make pm2 start automatically on server restarts let’s run:</p><pre>pm2 save<br>pm2 startup</pre><p>To check your server is running you can do pm2 status , you can read the docs on <a href="https://pm2.keymetrics.io/docs/usage/quick-start/">pm2 here</a></p><p>If you’ve made it this far you are a superstar ⭐️</p><h3>Integration with Github Actions</h3><p>This is the last step in our process. We’re going to create a .yml file for Github actions to update our server anytime a change is pushed to our master branch</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*wM9TJApFUH8J-x1flwrTBg.png" /></figure><p>In your repository, go to the ‘Actions’ tab and create a new custom workflow. I am calling mine nodejs.yml</p><p>Here is the .yml file for updating the repo and restarting pm2:</p><pre>name: Node.js CI/CD on Amazon Linux<br><br>on:<br>  push:<br>    branches: [ master ]<br><br>jobs:<br>  deploy:<br>    runs-on: ubuntu-latest<br><br>    steps:<br>    - uses: actions/checkout@v2<br><br>    - name: Install dependencies<br>      run: npm install<br><br>    - name: Deploy to AWS EC2<br>      uses: appleboy/scp-action@master<br>      with:<br>        host: ${{ secrets.REMOTE_HOST }}<br>        username: ${{ secrets.REMOTE_USER }}<br>        key: ${{ secrets.SSH_PRIVATE_KEY }}<br>        source: &quot;./&quot;  # This now correctly points to the current directory<br>        target: &quot;/var/www/your_project_directory&quot;<br><br>    - name: Restart PM2 application<br>      uses: appleboy/ssh-action@master<br>      with:<br>        host: ${{ secrets.REMOTE_HOST }}<br>        username: ${{ secrets.REMOTE_USER }}<br>        key: ${{ secrets.SSH_PRIVATE_KEY }}<br>        script: |<br>          cd /var/www/your_project_directory<br>          pm2 reload all  # This command restarts your application</pre><p>Make sure to update ‘your_project_directory’</p><p>Lastly, you’ll need to add these environment secrets REMOTE_HOSTthis is the ip address of your server, REMOTE_USER this is ec2-user for Amazon Linux 2 AMI, and then paste the contents of your .pem file as the SSH_PRIVATE_KEY You can do all of this in the repository settings:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*r8FWb7awKp0yqIMCLXMIeA.png" /></figure><p>That’s it! We’re done yay 😁</p><p>You can test this all by changing the ‘Hello World’ message and push that change to your master branch, then you should see the update at http://your.ip.address:3000</p><p>It is a bit of a process to set up, but when you’re developing it’s really nice to have an easy CI/CD pipeline automatically deploying changes to your REST API server. That way you can quickly deploy the changes live and see them in the frontend of your web / mobile application.</p><p>Let me know if this was useful in the comments and thanks for reading!!</p><p>Peace out,</p><p>— Wes</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f1f9a94bf4e2" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Not Just Coding: Must-Have Skills for Today’s Software Developers]]></title>
            <link>https://wesleybaxterhuber.medium.com/not-just-coding-must-have-skills-for-todays-software-developers-681e9f6519f0?source=rss-cb71643b03ea------2</link>
            <guid isPermaLink="false">https://medium.com/p/681e9f6519f0</guid>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[programming]]></category>
            <category><![CDATA[software-development]]></category>
            <dc:creator><![CDATA[Wesley Huber]]></dc:creator>
            <pubDate>Sat, 30 Dec 2023 16:27:14 GMT</pubDate>
            <atom:updated>2024-02-22T12:26:20.827Z</atom:updated>
            <content:encoded><![CDATA[<p>Software development is getting a lot more competitive. With the rise of AI tools, it’s easier than ever for anyone to be good at coding.</p><p>It might also seem tough to stand out when most programming jobs these days have over 1,000 applicants.</p><p>Here are some basic principals in which most software developers fall short.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZbGzgOTCMCuHcTWPF696MA.png" /></figure><h3>Communication</h3><p>It has long been a stigma that programmers are <a href="https://hackernoon.com/why-communication-is-important-for-software-developers">notoriously bad</a> at communicating. If you can respond to messages timely, speak clearly and professionally, and always keep your word — then you will set yourself apart from most software devs.</p><h3>Design, Structure, &amp; Standards</h3><p>Don’t just hack things together and make “spaghetti code” — be aware of software design best practices like <em>Object Oriented Design</em> and the <em>Model View Controller</em>.</p><p>When working in any project, structure and standards are key for long-term success. If you’re using a library or framework like React or Angular for example, you should have designated directories for your styles, your pages, and your components.</p><p>I recently worked on a React project where the structure was very disorganized and the page components were built like spaghetti code without <a href="https://bootcamp.uxdesign.cc/understanding-the-importance-of-separation-%EF%B8%8F-in-containerization-%EF%B8%8Fof-react-components-c2a8fa7f6e09"><strong>separation of concerns</strong></a>. Separating out API calls and reusable components and suggesting standards for our team to practice went a long way for our whole team.</p><h3>Meeting Deadlines</h3><p>I’m not sure if it’s because everyone seems to underestimate how long a task will take, or that they don’t account for testing and debugging.</p><p>In my 10 years of experience as a software developer, I’ve found it <a href="https://softwareengineering.stackexchange.com/questions/206475/are-missed-deadlines-common-in-programming-jobs">common practice</a> to miss deadlines. From independent freelance projects, to corporate SCRUM teams — freelancers and teams alike commonly fall short on their project objectives for releasing software.</p><p>If you can make accurate estimates and meet deadlines, and then you build a reputation to back that up, you’re a huge step ahead of the competition.</p><h3>Don’t Procrastinate — Be Proactive</h3><p>It’s easy when you’re in software development to procrastinate. You see a task, and you know what it will take to complete said task. Maybe you even over estimated the hours it would take, and you think it will actually take you a shorter time — so you procrastinate. This is easy to fall into, but it is bad practice.</p><p>Be proactive on your tasks. Think of it like a game where you get more points by going under the estimated hours and delivery time by as much as possible. That will score you the most points and set you apart from the average Software Developer.</p><h3>Stay in scope — do not Over Promise, Under Deliver</h3><p>Like missing deadlines, it’s common especially for new developers, to over promise and fall short of the targeted goal.</p><p>On the flip side of this, if you can manage to <strong>under promise<em> </em></strong>and <strong>over deliver</strong>, that will really set you apart from other software devs and your client will love you for it.</p><p>That being said, it is important to push back on certain things that are out of the agreed upon project scope. It’s a good idea to allow a bucket of hours for revisions, if the client is still not satisfied, then you request an adjustment of the contract. It’s also important to remain flexible, and the client will respect staying true to the contract. This is where communicating early, effectively, and often comes into play.</p><h3>Don’t give up — complete the project</h3><p>With all that being said, it’s important to not give up on a project. Work with your client or team to figure out what it means for said project to be completed. Do the very best to meet that goal, and make sure that both you and the client are satisfied with the end product.</p><h3>Optimization &amp; Efficiency</h3><p>There are always ways that things can be done better. Look for ways that the code can be improved or optimized, voice these early and often to make your client and team aware of potential optimization.</p><p>This is the secret sauce, if you can find a way to optimize software and show data to prove efficiency, you are GOATed. The goal of software is to optimize efficiency of rigorous tasks, if you can find a way to optimize or make something more efficient, you have stumbled upon the holy grail of programming.</p><h3>Be Kind, don’t be cocky</h3><p>It’s so important to be kind. I’ve been in a team where I got yelled at by the tech lead for not being able to get something the work the way he thought it should work. He then got on a call with me for over 2 hours and he couldn’t figure out how to make it work!</p><p>It made me so frustrated, but I did my best to not to retaliate. If the tech lead had been more kind from the beginning it would have been a better experience for everyone.</p><p>There are also situations where you might be more knowledgeable than your lead developer. In my own experience a lead developer on a React project did not know what was wrong with his repository when he simply needed to install dependencies from package.json — in situations like this it’s important to check yourself and not come off as arrogant, and respect that team members come from different backgrounds of experience.</p><p>You can always learn from others, even though I might be more experienced in React, the lead developer introduced me to the term “<a href="https://stackoverflow.com/questions/456177/function-overloading-in-javascript-best-practices">Function Overloading</a>”</p><p>It’s best to lead by example with respect and kindness. Programming should be about community and sharing knowledge. Not about shaming someone who is learning.</p><h3>Documentation</h3><p>Going along with the sense of community and learning, it’s important to document structure, standards, and your project’s codebase so future developers can easily be onboarded and everyone is on the same page.</p><p>Sometimes you might be facing a project like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/750/1*JQbbI6FZvZQqIcXKdu3FCg.jpeg" /></figure><p>It’s also good to keep in mind that any documentation is better than none!</p><h3>Aside from the humor &amp; while much of this article may seem obvious —</h3><p>Most software developers, myself included, are guilty of falling short on many of these skills. How would you rate yourself on these skills (1–10)?</p><p><strong>Communication:</strong></p><p><strong>Design, Structure, Standards:</strong></p><p><strong>Meeting Deadlines:</strong></p><p><strong>Not Procrastinating:</strong></p><p><strong>Staying in Scope:</strong></p><p><strong>Completing the project:</strong></p><p><strong>Optimizing &amp; making code the most efficient:</strong></p><p><strong>Being Kind:</strong></p><p><strong>Documentation:</strong></p><p>Take your lowest scores and focus on improving those areas this year!</p><p>Nobody is perfect, but if you can optimize these skills in your software development work, you will be successful.</p><p>Thanks for reading and happy coding!</p><p>Peace out,</p><p>— Wes</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=681e9f6519f0" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Beyond JavaScript: WebAssembly in Modern Web Development]]></title>
            <link>https://wesleybaxterhuber.medium.com/beyond-javascript-webassembly-in-modern-web-development-a83163b65109?source=rss-cb71643b03ea------2</link>
            <guid isPermaLink="false">https://medium.com/p/a83163b65109</guid>
            <category><![CDATA[webassembly]]></category>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[programming]]></category>
            <dc:creator><![CDATA[Wesley Huber]]></dc:creator>
            <pubDate>Tue, 12 Dec 2023 13:24:09 GMT</pubDate>
            <atom:updated>2023-12-12T13:24:09.041Z</atom:updated>
            <content:encoded><![CDATA[<p>Web Assembly or WASM, has been revolutionizing the web since its adoption by W3C in 2019, and it’s only going to keep shaking things up.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3a-Dhscifjs74JuKHG1IZQ.png" /></figure><p><a href="https://webassembly.github.io/spec/core/intro/introduction.html">What is Web Assembly</a>, or WASM?</p><blockquote>“WebAssembly (abbreviated Wasm <a href="https://webassembly.github.io/spec/core/intro/introduction.html#wasm">[1]</a>) is a <em>safe, portable, low-level code format</em> designed for efficient execution and compact representation.”</blockquote><p>Web Assembly has two formats — the first is .wat which is a low level assembly language. You use this .wat as the target language for native code written in C++, rust, Go, Java, Python, e.t.c. to compile.</p><p>Here’s an example of a .wat file that defines a function which adds two numbers:</p><pre>(module<br>  (func $add (param $a i32) (param $b i32) (result i32)<br>    get_local $a<br>    get_local $b<br>    i32.add)<br>  (export &quot;add&quot; (func $add))<br>)</pre><p>The .wasm file is a binary format, which is not human-readable. It&#39;s what you get after compiling the .wat file. Here&#39;s what the binary content might look like:</p><pre>00 61 73 6D 01 00 00 00 01 07 01 60 02 7F 7F 01 7F<br>03 02 01 00 07 07 01 03 61 64 64 00 00 0A 09 01 07<br>00 20 00 20 01 6A 0B</pre><h3>Why should you care?</h3><blockquote>If WASM+WASI existed in 2008, we wouldn’t have needed to create Docker. That’s how important it is. WebAssembly on the server is the future of computing. -<strong>Docker co-founder Solomon Hykes</strong></blockquote><h3>What can you build with WASM?</h3><ol><li>How about an in-browser N64 Emulator?<a href="https://github.com/nbarkhina/N64Wasm">https://github.com/nbarkhina/N64Wasm</a></li><li>Figma has created their editor using C++ and WebAssembly, <a href="https://www.figma.com/blog/webassembly-cut-figmas-load-time-by-3x/">cutting their load times by 3x</a></li><li><strong><em>Ethereum </em></strong>has proposed to redesign it’s Smart Contracts to perform better and reduce network latency and transaction fees using <a href="https://ewasm.readthedocs.io/en/mkdocs/">eWASM</a></li><li>AutoCAD was brought to the web <a href="https://madewithwebassembly.com/showcase/autocad/">thanks to WebAssembly</a></li></ol><p>If you’re a web developer, you don’t want to miss out on the benefits of WebAssembly! You can get started learning here: <a href="https://webassembly.org/getting-started/developers-guide/">https://webassembly.org/getting-started/developers-guide/</a></p><p>Peace!</p><p>-Wes</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a83163b65109" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>