Realistic evaluations of AI agents for software engineering.
GPT-6 Astra is here + safeguards.
SWE-Bench Pro has been accepted to ICML 2026. See you in Seoul!
I'm currently a Research Program Manager at OpenAI. My primary focus is on building safety systems, mitigating existential risk, and doing no harm.
Outside of work, I enjoy tennis, guitar, and German shepherds.
Realistic evaluations of AI agents for software engineering.
An open-source project enabling ML on proprietary radar signals.
Stanford University
MS in Electrical Engineering
University of Illinois at Urbana-Champaign
BS in Electrical Engineering
Frontier AI evaluations and products
Founding an early-stage company
Computer vision & robotics research
Computer vision in healthcare research