I build AI agent orchestration systems: multi-agent architectures that plan, execute, and review real work. Orchestration is a systems problem, and most teams treat it like a prompting problem.
That take comes from a decade of Python: frameworks, services, and the infrastructure underneath them, rebuilt until they were right. I'm building agent orchestration products on that thesis now, and writing it all down at zech.sh as I go.
I also run Smarter Dev, a 13,000-member developer community helping people build real skills in the age of AI.
If you need agents doing dependable work in production, not demos, I take on scoped consulting engagements: zech.sh




