All Posts

My real production workflow for shipping features with AI agents: 3 terminals, PID-like feedback loops, independent review, and quality gates. No vibe coding—just controlled engineering that serves real users.

Looking back at my 10 predictions for 2025 to see what came true, what missed the mark, and why

Highlights from my first PyCon talk, the magic of the community, the probabilistic nature of AI, and a tour of the multi-agent repo I demoed.

A concise primer on Docker Model Runner—the new beta in Docker Desktop that lets you pull, cache, and serve LLMs straight from Docker Hub through an OpenAI-compatible API.

MCP vs ACP: AI Protocols for Context and Agent Orchestration