About Me

I build AI systems and automations that need to work after demo day — RAG pipelines, agent workflows, and internal tools people actually open on a Monday. LangChain, LangGraph, FastAPI, and a lot of debugging in between.

At 1POINT1 I work on NL-to-SQL and document pipelines; before that, GenAI and compliance tooling at Cyber Security Umbrella.

That stretch — shipping under compliance pressure, then production AI — taught me to care about evals, guardrails, and the boring parts that keep systems running. I explore tools hands-on before asking a team to adopt them, and I'd rather ship something small that works than demo something big that doesn't.

Shipping taught me that getting retrieval right matters more than swapping models — chunk size, evals, and knowing when RAG is the wrong tool changed how I build and how I talk to teams about what's realistic.

I like owning problems end to end: schema design, API contracts, the prompt that breaks at 2am, and the dashboard someone actually opens. Early enough in my career to stay hands-on, far enough in to know when not to over-engineer.

Outside work I read papers I half understand, break side projects, and follow how teams are actually adopting agents — not just the launch tweets. Composition and curiosity from photography and music still show up in how I think about interfaces and flow.

Lately I've been exploring what MCP and tool-calling mean for real internal workflows — not as an expert, but as someone figuring it out. If you're in the same boat, happy to think through it together.