AI Project Manager. AI/ML engineer by training.
I work on the system around AI output: who owns it, how it is evaluated, where a human reviews it, and what is allowed to ship.
AI draft → engineer review → evaluation gate → human decision → ship
- No unmeasured claims. Evaluation should match the cost of failure.
- Human ownership stays explicit. A model can produce output. A person owns the decision.
- Architecture follows risk. Sometimes the right AI system deliberately excludes GenAI.

