jobs
Staff Research Scientist, Gemini Research, DeepMind
Just as grading olympiad proofs takes someone who can solve olympiad problems, judging an AI-proposed change to a frontier training algorithm takes a researcher who could have invented it.
- Company
- Location
- Mountain View, CA, USA
- Status
- Closed
- Posted
- 2026-10-14 09:08 UTC
A staff-level individual contributor on Google DeepMind's Gemini Research team, on site in Mountain View. The researcher invents ML architectures, optimizers and training methods, and builds automated systems in which Gemini proposes, codes and evaluates changes to its own training stack. The role also serves as final verifier of AI-proposed changes, reproducing results and running ablations across model scales. The person also builds training code in JAX for TPUs. The posting asks for a PhD or equivalent experience, four years of research work, publications and distributed deep-learning framework experience.
Original job posting