Research Scientist, Robotics RL, DeepMind
Research Scientist, Robotics RL, DeepMind Overview
| Company Name | |
| Job Role | Research Scientist, Robotics RL, DeepMind |
| Qualifications | Bachelor’s |
| Category | IT Jobs |
| Job Type | Full Time |
| Location | London |
DeepMind, a part of Google, is seeking a highly skilled Research Scientist specializing in Robotics Reinforcement Learning (RL) to join their team in London, UK. This role involves leading innovative research in developing algorithms and models for embodied AI and the pursuit of Artificial General Intelligence, with a focus on physical robotic systems. The ideal candidate will have a strong background in scalable machine learning techniques, particularly reinforcement learning, and experience working with real robots and simulators.
Key Responsibilities
- Design, develop, and evaluate algorithms that enable robotic agents to improve themselves and master tasks through post-training techniques.
- Lead research efforts in creating novel models for embodied AI, integrating large foundation models such as Gemini Robotics.
- Work collaboratively to invent new algorithms and push the boundaries of current robotic capabilities.
- Create prototypes and work directly with physical robots in lab and real-world settings to address practical use cases.
- Implement approaches for mid- and post-training of foundation models, including reinforcement learning and self-improvement techniques.
- Train robot policies using reinforcement learning in both simulation and real-world environments.
- Develop and optimize multimodal models, including vision-language and video models, for robotic applications.
- Build robust research pipelines and software to facilitate rapid hypothesis testing and experimentation.
- Set research goals, design experiments, analyze data, and communicate findings effectively within and outside the team.
Qualifications and Experience
- PhD in a relevant technical field or equivalent practical experience.
- At least two years of experience working with reinforcement learning or related techniques for training foundation models and robotic self-improvement.
- Hands-on experience with simulators and physical robotic systems, especially involving dexterous manipulation and whole-body control.
- Familiarity with multimodal sensing technologies such as tactile sensors.
- Proven ability to deploy models and algorithms on large-scale systems handling extensive real-world data.
- A passion for translating research into real-world robotic solutions.
- Experience with training approaches for foundation models, including reinforcement learning for agents to self-improve.
- Knowledge of large multimodal models like LLMs, VLMs, and video models, especially in the context of robotic agents.
About DeepMind
DeepMind is a pioneering AI research lab dedicated to advancing artificial intelligence to address complex global challenges and accelerate product innovation. The organization emphasizes safety and ethics in AI development and aims to use its technologies for widespread societal benefit. Working at DeepMind offers opportunities to contribute to cutting-edge research with a global, interdisciplinary team.
Additional Information
DeepMind is an equal opportunity employer committed to diversity and inclusion. They do not specify sponsorship options in the job posting, and the role is based in London, UK, with onsite work expected. Applicants requiring accommodations are encouraged to contact the company through their designated forms. English proficiency is required for all roles to facilitate global collaboration.
Degree Requirement: Bachelor’s
Visa Sponsorship May be
To apply for this job please visit www.google.com.