I build evaluation tools and interaction systems for agentic AI products, with a focus on turning ambiguous behavior into traces, scoring criteria, reproducible artifacts, and shipped interfaces.
Current focus: evaluation construction, agentic workflows, interaction design, product engineering, and instrumented human-AI collaboration.
Recent work: Dexterous, typed and inspectable scoring programs for interactivity-eval construction.
Previously: co-founded Nomikos/Sarmiza, built spatial and multimodal interaction prototypes, and shipped product/design systems across mobile and web.
Location: New York

