AI Engineer building production GenAI systems with evals, types, and human-in-the-loop guardrails.
Enterprise IT Systems Engineer (GCP, Workato, Workspace) transitioning full-time into AI engineering. I ship typed Python, instrumented pipelines, and agents that earn their autonomy.
Core AI/ML
Backend
-261230?style=flat-square){kind=icon}
Infra & Ops
Eval-first. Every GenAI project ships with an eval harness before it ships a feature. If you can't measure it, you can't trust it in production.
Typed, modular Python. mypy --strict, Ruff, no notebooks in main, no untyped public APIs. Code that's read more than it's run.
Human-in-the-loop on irreversible actions. Models propose; engineers commit. Reversible architecture by default. Every action has an undo path.
Observable by construction. Langfuse traces, structured logs, cost + latency budgets. Debugging a production agent at 2am should not require new instrumentation.