Research Scientist, Robotics Pre-Training & Data Quality

Salary: £160,000 - 200,000 per year

Requirements:

• PhD in a technical field or equivalent practical experience
• Experience with algorithmic architectures, data sources, and training and inference techniques for generative multimodal models, especially robotics applications such as VLAs and WAMs
• Experience optimizing models and systems through performance tuning, experimentation, and debugging
• Practical experience iteratively improving training recipes and model performance against benchmarks
• Experience with simulators and real-world robotic platforms, such as dexterous manipulation and multimodal sensing
• Experience with benchmarking, data sources, data labeling strategies, and data mixture optimization for robotics foundation models
• Ability to build and maintain software tools for rapid research iteration
• Experience with robot policy training using imitation learning and related techniques, and with the practical considerations of pre-training
• Experience managing large volumes of multimodal data and optimizing data mixtures for general, non-robotics frontier models, including LLMs, VLMs, image and video generation, and world models
• Passion for bringing research from the lab to robust, real-world robotic systems and solving real-world robotics problems

Responsibilities:

• Drive data quality, acquisition, and labeling strategies, and oversee mixture optimization for robotics foundation models
• Coordinate training runs, improve training recipes, and systematically improve model performance against benchmarks
• Build and maintain software tools and processes that enable rapid research iteration and verification of data quality through robot policy training
• Contribute to research in reinforcement learning, vision-language-action modeling, world-action models, and simulation
• Set up large-scale tests and quickly deploy promising ideas
• Manage research deadlines and deliverables while developing new or improved products, processes, and technologies
• Create experiments, prototype implementations, and design new architectures
• Share and publish findings, including research inspired by internal projects and collaborations with external research programs

Technologies:

• AI
• Support
• Model Training
• Robotics
• Hardware
• Mobile

More:

We are Google DeepMind, a pioneering AI lab whose interdisciplinary teams advance AI to address complex global challenges and support product innovation for billions of users. We work toward widespread public benefit and scientific discovery, with safety and ethics as our highest priorities. Our global teams offer learning opportunities and varied career pathways, and our research spans fundamental work, product innovation, product contributions, and infrastructure. We provide individuals and teams with flexibility to emphasize different types of work. This Research Scientist role is for someone eager to apply foundation model training to real-world robotics problems, working in a fast-paced, collaborative environment and contributing to research with practical impact.

last updated 39 week of 2026