Research Scientist - Agentic Reasoning & RL Systems

DeepMind Technologies Limited; Mountain View (CA); On-site; USD 207,000 - 300,000; Full time

Benefits offered by this job

Equity

Bonus target

Benefits package

Job summary

DeepMind Technologies Limited invites applications for a senior researcher/engineer on the PRISM team. You will work across the full research and engineering lifecycle, developing distributed post-training infrastructure and algorithms that enable Gemini models to solve complex multistep problems autonomously.

We pursue breakthroughs and production-ready capabilities for Gemini and Gemma, with a focus on scalable reasoning, RL scaling, and responsible AI.

Qualifications

• Bachelor's or Master's in a quantitative field with practical experience.
• 4+ years building and scaling ML models with deep learning frameworks.
• Experience in RL, SFT/RLHF/RLAIF, or related agentic areas.

Responsibilities

• Operate across the full research-and-engineering lifecycle of frontier reasoning and agentic systems.
• Address unsolved problems in agentic reasoning, turning early prototypes into hardened production features for Gemini releases.
• Architect and optimize distributed post-training pipelines and agent-environment simulation loops across thousands of accelerators.
• Design rigorous experiments and failure analyses to isolate performance bottlenecks and communicate findings.
• Drive technical excellence by maintaining high code quality and architectural health across shared reinforcement learning and modeling codebases.

Skills

ML frameworks

Reinforcement Learning

Distributed systems

Python

Education

Bachelor's/Master's in CS/Math/Physics

Tools

JAX

PyTorch

TensorFlow

Job description

DeepMind Technologies Limited invites applications for a senior researcher/engineer on the PRISM team. You will work across the full research and engineering lifecycle, developing distributed post-training infrastructure and algorithms that enable Gemini models to solve complex multistep problems autonomously.

We pursue breakthroughs and production-ready capabilities for Gemini and Gemma, with a focus on scalable reasoning, RL scaling, and responsible AI.