About

I build evaluation and systems infrastructure for tool-using agents. At Google, I work on agent framework efficiency and quality, with a focus on making agent systems faster, more reliable, and less expensive. I'm especially interested in agent evaluation, ML inference, GPU systems, compilers, and runtimes.

I also publish reproducible studies on agent evaluation and reliable software. My recent work on semantic state diffs tests an evaluation layer for the external-state changes claimed by long-horizon agents. It reports controlled simulator results, not observed agent performance. I founded Strust, an independent benchmark for code conversion and conformance in legacy modernization.

Outside of work I play cello, backpack, write, and release music as Copperbelt.

Elsewhere: software, social, research / projects, writing, music, photos.

Current and recent

• I work on agent framework efficiency and quality at Google Cloud AI.
• I published a controlled study of semantic state diffs for evaluating external state changes in long-horizon agents.
• I founded Strust to measure code-conversion performance and conformance independently.
• I released jazz and blues recordings as Copperbelt.

Education

• Columbia University: MS in Machine Learning and Bioinformatics.
• UC Santa Barbara: BS in Chemical Engineering.
• UC Santa Barbara: BS in Statistics and Data Science.

Selected work

Current work and collaborations:

• Google Cloud AI (agent framework efficiency and quality)
• Semantic state diffs (controlled evaluation of claimed external-state changes)
• Strust (code-conversion benchmarks and conformance metrics)
• Copperbelt (jazz/blues releases)