Roy Frostig
Scholar

Roy Frostig

Google Scholar ID: UoATnWEAAAAJ
Research scientist, Google DeepMind
machine learningoptimizationdifferentiable programminglanguagessystems
Citations & Impact
All-time
Citations
11,007
 
H-index
15
 
i10-index
17
 
Publications
20
 
Co-authors
14
list available
Publications
20 items
Browse publications on Google Scholar (top-right) ↗
Resume (English only)
Academic Achievements
  • Published numerous papers in top-tier conferences such as ICML, NeurIPS, ICLR, including but not limited to 'Gemini 2.5: Pushing the frontier with advanced reasoning, multimodality, long context, and next generation agentic capabilities', 'You only linearize once: Tangents transpose to gradients', etc.
Research Experience
  • Research scientist at Google DeepMind; co-author and core developer of JAX.
Education
  • PhD from Stanford University, advised by Percy Liang.
Background
  • Research interests: foundations of machine learning, systems, and programming languages. Worked in the statistical machine learning group at Stanford.
Miscellany
  • Contributed to the early stages of the Rust programming language.