Technical Program Manager, Scientific Intelligence for Gemini, DeepMind
Mountain View, CA, USA · Manager · Full-time · Posted Oct 3, 2026
- Experience
- 5 years of experience with technical programs, research engineering, or applied AI/ML projects (or 3 years with a PhD).
- Education
- Bachelor's degree in a Life Science (e.g., Computational Biology, Chemistry), Physical Science (e.g., Physics, Materials Science), Mathematics, or Computer Science, or equivalent practical experience.
About the job
Leads scientific reasoning program execution for Gemini, turning research hypotheses into training data, evaluations, verifiable reinforcement-learning environments, and model-release milestones.
What you'll do
- Build roadmaps and release milestones with science and reasoning research leads.
- Coordinate data collection and generation, supervised fine-tuning, verifiable RL training environments, and evaluation work.
- Analyze model checkpoints and reasoning traces, address capability gaps, and manage cross-functional dependencies.
What they ask for
- Bachelor's degree in a Life Science, Physical Science, Mathematics, or Computer Science, or equivalent practical experience.
- 5 years of experience with technical programs, research engineering, or applied AI/ML projects (or 3 years with a PhD).
Nice to have
- Experience analyzing model reasoning traces on graduate-level science problems and translating identified issues into training changes.
- Knowledge of SFT data quality, RL reward signals, verifier calibration, and capability tradeoffs across model sizes; familiarity with Python-based scientific computing stacks.
Pay
$217000 - $236000 USD base salary
Tools
Gemini, SFT, RL, Python
AI focus
Advances scientific reasoning and agentic capabilities in Gemini through training data, verifiable RL environments, evaluations, and model development.
Model-written summary · the original posting is authoritative