My thoughts on frontend architecture, AI integration, and building things that scale.
Most tech organizations don't have an intelligence problem — they have a fragmentation problem. Product intent lives in tickets, design rationale lives in Figma, engineering context lives in PRs, and nobody's talking to each other. Here's what it looks like to fix that with a shared LLM reasoning layer instead of a departmental chatbot.
Read moreEngineering teams have spent decades learning to estimate time. The next skill they need to master is estimating tokens — because AI features have a cost dimension that doesn't behave like anything they've budgeted before.
Read moreThe industry is moving fast to reward engineers who can prompt and deliver. But speed without comprehension isn't engineering — it's a liability waiting to be discovered. Here's what gets lost when we stop reading what we ship.
Read morePrompt engineering is reshaping how frontend engineers work — faster scaffolding, smarter tooling, less boilerplate. But speed without governance is just risk in disguise. Here's an honest look at both sides.
Read moreHow I architected a scalable AI microfrontend ecosystem integrating LLM services like Claude and OpenAI across independently deployable UI domains — and what I learned along the way.
Read moreThe best senior engineers I've worked with all had one thing in common — they were the first to say they didn't know. Here's why humility isn't a soft skill, it's a technical one.
Read moreA full walkthrough of how I wired together Simli, Claude, Supabase, OpenAI TTS, and my own Instagram export data to build an AI avatar that actually sounds like me — and the surprisingly tricky engineering it took to make it feel seamless.
Read moreI've taught full stack development at two different bootcamps, and it changed how I work, how I communicate, and how I think about this industry. Every working developer should do it at least once.
Read moreMulti-user loan application workflows are inherently complex. At Spring EQ, we used a microfrontend architecture to deliver a scalable co-borrower flow that could evolve independently of the rest of the platform.
Read moreEight production SPAs. Bloated bundles pushing 6MB. Build times that killed developer momentum. Here's how we diagnosed the problem, made the call to migrate to Vite, and shipped all eight apps with zero downtime.
Read moreComponent libraries are easy to build and easy to abandon. At Guild Mortgage, we built one that 8 teams actually adopted — and it saved over 125 developer hours every month. Here's what made the difference.
Read moreThe most impactful software I've shipped wasn't a customer-facing product. It was an internal tool at Guild Mortgage that quietly gave people thirty hours of their week back.
Read moreWorking inside Amazon on AWS OpenSearch meant operating at a scale where every UI decision had real downstream consequences. Here's what that environment taught me about building interfaces that hold up.
Read moreThe question was whether Instagram growth could be automated intelligently — not with spam bots, but with something that actually behaved like a thoughtful, engaged user. The answer turned out to be yes.
Read moreWhen Allegiant Air transitioned from waterfall to Agile, the hardest part wasn't the process — it was getting everyone speaking the same language. So I built them a patient, always-on teammate powered by IBM Watson.
Read moreWhen the NBA needed a mobile ticketing solution for their All-Star weekend in Chicago, the requirement was simple: track every transaction live and give stakeholders a real-time picture of ROI as it happened.
Read moreThe pressure to ship faster is real, legitimate, and not going away. But somewhere between agility and chaos, the industry has started treating discovery, grooming, and acceptance criteria as optional. They're not. Here's how to protect what matters without slowing your team to a crawl.
Read more
Marcus Bass
AI · Career Q&A