About

I work where perception research meets software architecture: enough depth to reason about the model, enough breadth to ship the system around it.

Most of my time goes to problems where a model alone is not the answer — where the data path, the latency budget, the alerting behavior, and the interface all decide whether the result is actually useful.

I care about work that leaves evidence. Architecture that can be explained, decisions that record their rejected alternatives, and results that can be measured rather than asserted.