Neel Nanda
Runs the mechanistic interpretability team · Google DeepMind
At a glance. Neel Nanda is listed as Runs the mechanistic interpretability team at Google DeepMind. The profile sits among the builders and open-source contributors. Fields listed: Interpretability, Safety and alignment and Developer tools. The source lists 2 official pages and open-role pages for Anthropic.
Summary
Runs the Google DeepMind mechanistic interpretability team; creator of the TransformerLens library.
Known for: TransformerLens; Gemma Scope: Open Sparse Autoencoders Everywhere All At Once on Gemma 2; Progress measures for grokking via mechanistic interpretability.
Works at
- Google DeepMind 64 people in the wiki
Other documented roles
Not confirmed as current in the source profile.
- Language model interpretability researcher, Anthropic
Open roles
Related people
- Arthur ConmyAlso at Google DeepMind
- Adam GleaveAlso at Google DeepMind
- Shane LeggAlso at Google DeepMind
- Philipp SchmidAlso at Google DeepMind
- Anca DraganAlso at Google DeepMind
- Kenneth LiInterpretability · Safety and alignment
- Kush R. VarshneySafety and alignment · Interpretability
- Harold BoothSafety and alignment · Developer tools
- Chris OlahInterpretability · Safety and alignment
- Duen Horng "Polo" ChauInterpretability · Safety and alignment
Source and corrections
Read the full profile on USASI ↗
Facts on this page come from that USASI profile (reviewed Oct 21, 2026); data last checked Oct 23, 2026. See something wrong? Suggest a correction.