Every PR on my team gets reviewed by Claude before a human sees it. Here's the exact prompt and CI setup.
AI code review isn't about finding bugs—static analysis does that better. It's about pedagogical review: explaining trade-offs, questioning assumptions, and raising standards without slowing the team down.
We give Claude the full PR diff plus the codebase context. We ask it to review for readability, maintainability, and whether the change leaves the codebase in a better state. We ignore style—linters handle that.
The review runs as a GitHub check. It doesn't block merges; it educates. Engineers read the comment, decide if it's valid, and move forward. After a month, code quality visibly improves.
A voice agent that handles negotiations with real humans is not the same as a chatbot. Here's the architecture that keep...
RAG is not a silver bullet. Here's when it works, when it doesn't, and how to know which problem you're actually solving...
Every SaaS I build starts from this template. Auth, billing, database, and AI hooks pre-wired. Here's what's included an...
How to move from pre-sales engineering into an architecture role without taking a step backward.Pre-sales teaches you to...
A short, opinionated list of books that shaped how I think about systems.Most architecture books are verbose and outdate...
My exact dev stack. Updated quarterly.EditorVS Code with Copilot. The debugger integration alone is worth it.TerminalFis...
Papers, blog posts, and talks every AI-focused engineer should read.FoundationStart with Attention Is All You Need if yo...
Remote work, visa sponsorship, OSS opportunities, and the things I wish someone had told me ten years ago.Getting remote...