Notes on building AI
that actually ships.

How I built ai-dev — a multi-agent engineering system using Claude Opus and Claude Code that plans, implements, reviews, and iterates on software autonomously.

After deploying language models in healthcare environments, here are the hard lessons learned — from prompt drift to latency budgets to hallucination mitigation at scale.

Most enterprise AI pilots never make it to production. Here's a pattern I've seen repeatedly — and a framework for getting from proof-of-concept to deployed system.