Bilal Chughtai

London, UK

Bio

Updated 03/22/26

Bilal Chughtai is a Research Engineer on the language model interpretability team at Google DeepMind, where he has worked since February 2025 within the broader AGI safety and alignment team. He studied Mathematics at the University of Cambridge, completing his undergraduate degree and a Part III (MMath) in 2021. Before joining DeepMind, he conducted independent mechanistic interpretability research supported by the Long-Term Future Fund, including a project mentored by Prof. David Bau of Northeastern University. He was also a fellow at the MATS (ML Alignment Theory Scholars) program. His notable publications include "A Toy Model of Universality: Reverse Engineering How Networks Learn Group Operations" (ICML 2023, co-authored with Lawrence Chan and Neel Nanda) and "Open Problems in Mechanistic Interpretability" (2025, co-authored with Lee Sharkey, David Bau, and over 25 other researchers). His research focuses on understanding the internal mechanisms of neural networks to advance AI safety.