About me

I am a researcher based in London, United Kingdom. My career goal is to help understand AI systems, thereby reducing the risks they pose and increasing the probability of a prosperous future in their presence.

I’m a member of technical staff at Goodfire, developing and scaling fundamental interpretability methods.

Summary of my past professional experience:

• Founding member, Lead Engineer and Head of Security at Apollo Research. Helped bring Linear Parameter Decompositions to the field (APD). Also contributed to work on end-to-end SAEs.
• Research Engineer at Conjecture, working mostly on Mechanistic Interpretability (e.g. early work on sparse autoencoders, polytope lens).
• Worked on understanding the neural mechanisms of behaviour in reinforcement learning agents (Distill-style article).
• Research Engineer at NukkAI, developing logic-based AI systems and software tools to make and understand decisions in the game of Bridge.
• Honours in Computer Science at The University of Sydney, graduating with Honours Class I and The University Medal (WAM: 92.5). My thesis was on reinforcement learning with temporal logic objectives.
• Developed profitable predictive models for sports betting.
• Poker and bridge player.
• Earned a Bachelor’s of Mathematics from The University of Wollongong, Australia.

Other interests of mine include football (soccer), hiking, meditation, and finding ways to help conscious beings flourish.