AI Engineer · building agentic systems, LLM infrastructure & B2B AI products
- Agentic systems — multi-agent orchestration, tool-calling harnesses, evaluation loops
- LLM infrastructure — self-hosted gateways (LiteLLM), routing, caching, observability & cost control
- B2B AI products — compliance, diligence and workflow automation for real businesses
I care about systems that ship and measurably work: real users, real cost, real evals.
More about how I build
- Prefer boring, observable infrastructure over clever abstractions
- Instrument before optimizing — spend, latency and failure modes get measured first
- Ship in small, verifiable increments; treat evals as a first-class artifact
Always building. Open to conversations about agentic systems and AI infrastructure.


