Research Scientist, Robotics Data, Pretraining, DeepMind

What you'll do

• Develop data quality, filtering, labeling, and multimodal data-mixture methods for Gemini pre-training.
• Run pre-training and policy-training experiments, improve training recipes, and benchmark physical reasoning and action performance.
• Build research software and evaluation tools; assess data and robot policies through offline and closed-loop evaluations.

What they ask for

• PhD in Computer Science, a related technical field, or equivalent practical experience.
• Experience with robot policy learning, such as imitation learning or RL, across simulators and real-world robotic platforms.
• Experience working with simulators and robots.
• Research contribution experience, including publishing conference papers.

Nice to have

• Experience with benchmarking, multimodal data, labeling strategies, and data-mixture optimization for frontier or robotics foundation models.
• Experience iteratively improving model performance and building research software and tools.

Pay

$147000 - $210000 (USD) (base salary)

Tools

Gemini, RL, NeurIPS, CoRL, ICML, ICLR

AI focus

Research on integrating physical and action knowledge into Gemini pre-training and improving robot reasoning and action capabilities.