Writing
Technical writing about AI systems, full-stack engineering, and building products that scale.
Most RAG demos look great. Most RAG systems in production don't. Here's what I learned shipping several of them — chunking strategies, retrieval quality, re-ranking, and eval frameworks.
LangGraph makes it easy to build agentic systems, but the devil is in the details. Deep-dive on state management, error recovery, human-in-the-loop patterns, and streaming.
Performance, CodePush, Detox testing, and the architectural decisions that let us scale without rewriting everything. Lessons from the Ohi App.
Fonts, images, code splitting, caching — the full checklist of what actually moves the Lighthouse score needle, and what's just cargo-culting.