Reducing E-commerce Latency by 23 Seconds

Stripe formatted 25 million lines of code overnight. Not by rewriting it. Just formatting it. That sounds simple, but it took years of work, a zero-config Rust formatter, and a careful rollout across one of the largest Ruby codebases in the world. Stripe eventually got to 100% of its 42 million lines of Ruby being formatted with rubyfmt. And honestly, that is the kind of engineering I really respect. Not flashy. Not loud. Just making the whole system easier to work in. Because at scale, even “small” improvements become huge wins. Cleaner diffs. Fewer code review nitpicks. Less mental overhead. Better developer experience. The best tools are often the ones you stop noticing. They just work.

I’m not a software developer. But this year I’ve built our entire Amazon software, our lead gen system, and our audience segmenting for Google and Meta. The thing that made it possible isn’t talent. It’s design patterns. When I started using Claude Code, my code worked… until it didn’t. Add one more platform and the whole thing turned into spaghetti. Every new feature broke something else. Sound familiar? It’s the same problem I see in most ad accounts. So I did what an engineer does. I went back to first principles. I keep a single file on my computer with every design pattern in it. Adapter. Factory. Singleton. Strategy. Template. Now, every time my code gets hard to read, I don’t add another patch. I ask Claude Code to review that file and find a pattern that fits. The result? When I need to support a new platform, I copy its API doc into Claude Code, apply the right pattern, and it just slots in. No rewrite. No spaghetti. And this is the exact same skill that scales an ad account. You don’t manage a 100k-SKU account by reacting to every problem as it lands. You build a structure that absorbs new things without breaking. Adapters for the messy inputs. A strategy you can swap. Templates you reuse instead of rebuilding from scratch. Whether it’s code or campaigns, the lesson is the same: Don’t bolt on. Build something that scales. The tools are finally good enough that you don’t need to be an expert to build. You just need to think in structures. What’s the one thing you built this year that you never thought you could?

Two weeks already for this. Jeez. It's with no small amount of irony that the product I thought of is all about the UX. UX to me is sweet font on a solarized dark command line. No command lines here, but much more nuance in the few pages users see than it will appear at first glance. Or at least that's the goal.

A founder messaged me 3 months after I handed off their product. Not to say thank you. To say the new developer had been stuck for 2 weeks trying to figure out why payment webhooks kept failing in production. I had fixed that bug twice during the build. Never wrote it down. That message changed how I work. Now, before I close any project, I sit down and answer one question: what would the next developer Google at 2 am when something breaks? Then I document that. Not the obvious stuff. The obvious stuff lives in the code. The weird stuff. The decisions that made sense at the time and would look insane to anyone coming in cold. Why we used Supabase Realtime instead of polling. Why there is a second Stripe webhook handler that looks like a duplicate but is not. Why staging has a different auth flow than production. I built this into a full playbook from everything I have learned across 10+ MVPs. It covers: The environment map founders have never seen Every credential and where to rotate it Architectural decisions and why they were made Break/fix steps for the 5 most common failures A dual sign-off checklist for the engineer and the founder If you are a founder about to hand your product to a new developer, this is what your outgoing engineer should give you. Free. No email required. Swipe through below.

Most apps don’t crash. They choke slowly as users grow. • More traffic → slower performance • More features → messy codebase • More hires → slower execution That’s how scaling quietly breaks products. The smartest teams avoid this early. They build on 𝐬𝐭𝐚𝐜𝐤𝐬 𝐝𝐞𝐬𝐢𝐠𝐧𝐞𝐝 𝐟𝐨𝐫 𝐠𝐫𝐨𝐰𝐭𝐡 — 𝐥𝐢𝐤𝐞 𝐌𝐄𝐑𝐍. Because when your frontend, backend, and database speak the same language: • Shipping gets faster •Scaling gets simpler Teams stay aligned 𝐍𝐨 𝐩𝐚𝐭𝐜𝐡𝐰𝐨𝐫𝐤. 𝐍𝐨 𝐛𝐨𝐭𝐭𝐥𝐞𝐧𝐞𝐜𝐤𝐬. 𝐉𝐮𝐬𝐭 𝐦𝐨𝐦𝐞𝐧𝐭𝐮𝐦. If you’re serious about building something that actually scales, this is worth your time. https://lnkd.in/dvVHHT-G #MERN #TechStrategy #ScalableSystems #BuildToScale #StartupEngineering #ModernWeb #JavaScriptStack #EngineeringLeadership #SaaSDevelopment #SystemDesign #HiddenBrains"

Stripe’s Minions are redefining software engineering and every company will follow. This is the new blueprint for AI-native software engineering.

Hot take: micro-frontends are the most over-engineered thing we've shipped to frontend teams in years. Module Federation was built for Amazon-scale orgs. Somehow every 12-person startup is now using it. Worked with a team last month running 4 React versions on the same page. 2.1MB of duplicated runtime. Custom event bus to share state. Design system updates needed coordinated deploys across 4 repos. We moved them to a modular monolith with proper domain boundaries. Build went from 38s to 11s. Bundle dropped to 740KB. Half the bugs. Micro-frontends solve an org problem, not a code problem. If your team is under 50 engineers, you don't have the org problem yet. Genuinely curious what teams who scaled MFEs successfully would push back with. #modulefedration #mfes #microfrontend

Stripe's Minions agents ship 1,300+ pull requests a week. The story isn't the volume. It's the architecture, and what it says about every team that still calls the developer their bottleneck.

Just tried Bun for the first time. Wasn't expecting much. Was very wrong. 262M ops/sec vs Express at 15M. That gap isn't academic—it's the difference between scaling smoothly and hitting infrastructure limits. But what actually matters: Bun eliminates friction. No separate build pipeline. No TypeScript setup tax. No dependency on five different tools just to ship an API. It's purpose-built for modern development workflows. Express will always be foundational. It powers the backbone of the web. Stable, predictable, millions of companies depend on it. That's not changing. But here's the thing: when you actually build with Bun, you realize how much operational complexity we've normalized. Better runtime, better defaults, better developer experience. These compound into real business value—faster time to market, fewer DevOps headaches, cleaner codebases. Not suggesting a wholesale migration. But for teams evaluating their tech stack or starting something new? Bun deserves serious consideration. If you're exploring this space, curious what's working for your team. Using Bun in production yet? #webdevelopment #javascript #nodejs #backend #engineering

Did you know you can validate a credit card without making a single backend server call? 💳 It sounds like magic, but it’s actually pure math—specifically, Luhn’s Algorithm. Whether you’re booking a flight, shopping online, or building a checkout system, this clever 1954 algorithm is the first line of defense running instantly right in your browser. Here is exactly how it works in 5 simple steps: 1️⃣ Start backwards: Take the credit card number and look at the second-to-last digit. 2️⃣ Double up: Double every alternate digit moving from right to left. 3️⃣ Sum the digits: If doubling a number results in a two-digit number (like 12), add those two digits together to form a single digit (1 + 2 = 3). 4️⃣ Total it all: Add all the modified digits together with the remaining untouched digits of the card. 5️⃣ The Final Check: If the total sum is perfectly divisible by 10 (ends in a 0), the credit card number is valid! Why does this matter for Engineers and Product Builders? Performance and UX. By catching typos and invalid card inputs directly on the client side, you eliminate unnecessary API latency, save server bandwidth, and give users instant feedback before they even hit "Submit." Simple engineering tricks from decades ago are still quietly powering the modern web today. 🚀 👉 Have you ever implemented Luhn's Algorithm in your codebase, or do you rely on third-party libraries like Stripe to handle it all? Let me know in the comments! Credit to: take U forward Watch the breakdown here: https://lnkd.in/gv7-qCWK #SoftwareEngineering #WebDevelopment #Coding #Fintech #SystemDesign #LuhnsAlgorithm