About

I'm an AI Backend Engineer. I work on the parts of AI systems you never see in a demo — retrieval pipelines, agent orchestration, evaluation loops, and the glue between models and the products people actually use.

I came up through full-stack and design engineering, which still shapes how I build: with a bias toward clarity, small interfaces, and systems that behave well when the happy path breaks.

These days I work on adaptive learning engines, voice agents, and retrieval systems. LangGraph, FastAPI, Qdrant, and Python sit in the middle of most of my stack, with TypeScript and PostgreSQL at the edges.

Building AI from this side of the world means the infrastructure questions are never theoretical — latency, cost, and availability are load-bearing design inputs. I write about that, and about what breaks between a model that works in a notebook and one that works for a user on a flaky connection.

I care about user and developer experience, evaluation discipline, and not shipping systems I can't explain. I'd rather delete code than abstract around it, and I'd rather measure a pipeline than argue about it.