How I think about engineering.
Less biography, more of what I actually care about when I sit down to build.
I'm a software engineer first. Most of my work lives at the intersection of applied machine learning and the systems that put it into production — perception pipelines, retrieval and reasoning layers, evaluation harnesses, and the typed, testable software that keeps them honest.
My default lens is systems thinking. A model is one component; what surrounds it — data contracts, interfaces between layers, telemetry, deploy paths — decides whether it becomes a product or a demo. I try to design for the version of the system that exists six months in.
I care about evaluation before optimization, small honest interfaces, and the kind of software craftsmanship that compounds over the life of a codebase. I'd rather ship a small, calm system that quietly works than a flashy one that doesn't survive real load.
Principles
Production quality is the point.
Software that quietly works for years is worth more than software that demos well for a week. Ship things that hold up under real load.
Evaluation before optimization.
Especially in AI: if you can't measure a thing honestly, you can't improve it. Eval harnesses are the unit of progress, not throughput.
Systems, not scripts.
A model is one component. What surrounds it — retrieval, guardrails, telemetry, deploy paths — decides whether it's a product or a demo.
Craft compounds.
Taste, restraint, and honest interfaces cost the same as sloppiness in the moment and pay dividends for the life of the codebase.