Summary
Tom Bewley is an AI scientist focused on safety, currently at Mistral AI, with eight years' experience bridging cutting-edge research and industrial deployment. He holds a PhD from the University of Bristol on interpretable tree models for agents and has published at top venues (NeurIPS, AAMAS, ICRA) from work done at the Alan Turing Institute, Thales, and other labs. His background spans trustworthy AI, preference-based and interpretable reinforcement learning, and human-in-the-loop systems—skills honed through selective secondments and industry roles including J.P. Morgan’s Trustworthy AI COE. Tom combines deep research rigor with practical engineering experience (from CAD and FEA in early placements to production-focused ML safety), making him adept at turning theoretical advances into auditable, deployable solutions. An underappreciated strength is his track record of translating hands-on mechanical and systems engineering insights into more interpretable, robust AI designs.
8 years of coding experience
1 year of employment as a software developer
Altrincham Grammar School for Boys
Doctor of Philosophy - PhD, Tree Models for Interpretable Agents, Doctor of Philosophy - PhD, Tree Models for Interpretable Agents at University of Bristol
German, French, English