I led the software team at Vow, a scrappy startup turned world-leading cultivated meat company. I'm now a Manifund grantee pursuing research in AI safety, with an early focus on dangerous capability evaluation.
-
Measuring Attacker and Monitor Capability, Task Bias, and Attack Selection
26 Nov 2025 -
Extending the Cyber Task Horizon Study
26 Nov 2025 -
Scrapped Proposal: Measuring Deceptive Capability with Human Time Horizons and Control Protocols
08 Aug 2025 -
AI Task Length Horizons in Offensive Cybersecurity
02 Jul 2025 -
A Little Physics Sandbox
06 Mar 2025