Research Scientist Intern, Robotics Foundation Models Post-Training (PhD)

Meta; Zürich; On-site; CHF 31,000 - 47,000; Full time

Job summary

Meta Reality Labs Research is seeking a Research Scientist Intern to develop scalable offline reinforcement learning methods for post-training embodied AI models used in robotics. You will work with researchers and engineers to design, implement, and evaluate algorithms that push the frontier of world action models and vision-language-action models for robotic manipulation.

Internships range from 12 to 24 weeks with multiple start dates. Applicants should be pursuing a Ph.D.

Qualifications

• Currently pursuing a Ph.D. in CS/ML/AI/Robotics or related STEM field.
• Experience in reinforcement learning including policy/value/off-policy learning.
• Experience implementing and training deep learning models with Python and PyTorch.
• Experience designing controlled experiments, analyzing results, and debugging ML algorithms.
• Interpersonal experience across groups and cultures.
• Must have work authorization in country of employment and plan to return to degree program after internship.
• Experience developing offline RL methods for continuous control with large datasets.
• Experience training VLAs, WAMs, or other foundation models for embodied decision-making.
• Experience with diffusion or flow-based policies, action-sequence prediction, model-based RL, or world models.
• Hands-on experience with distributed training across GPUs/nodes.
• Experience evaluating learned policies on robotics benchmarks.
• Proven track record of significant results (grants, fellowships, papers) in top venues.

Responsibilities

• Develop scalable offline RL methods for post-training robotics models.
• Collaborate with researchers and engineers to build and evaluate model capabilities.
• Contribute to the next generation of embodied AI and robotics research.

Skills

Reinforcement learning

Offline RL

Python

PyTorch

Distributed training

Robotics

Experiment design

Research collaboration

Education

Ph.D. in Computer Science / ML / AI / Robotics

Tools

PyTorch

Job description

Meta Reality Labs Research is a team of researchers and engineers pushing the frontier of AI and robotics. Our team is building the next generation of world action models (WAMs) and vision-language-action models (VLAs) for robotic manipulation. We are seeking a Research Scientist Intern to develop scalable offline reinforcement learning methods for post-training these models.

Interns in this role will conduct fundamental and applied research to advance foundation models for robotics, with a focus on post-training. You will collaborate with research scientists and engineers to develop new algorithms, build and evaluate model capabilities, and contribute to the next generation of embodied AI.

Our internships are twelve (12) to twenty-four (24) weeks long and we have various start dates throughout the year.

• Currently has or is in the process of obtaining a Ph.D. degree in Computer Science, Machine Learning, Artificial Intelligence, Robotics, or a related STEM field
• Research experience in reinforcement learning, including policy optimization, value learning, or offline/off-policy learning
• Experience implementing and training deep learning models using Python and PyTorch or a comparable framework
• Experience designing controlled experiments, analyzing results, and debugging machine learning algorithms
• Interpersonal experience: cross-group and cross-culture collaboration
• Must obtain work authorization in country of employment at the time of hire and maintain ongoing work authorization during employment
• Intent to return to degree program after the completion of the internship/co-op
• Experience developing offline RL methods, particularly for continuous control, large datasets, or learning from suboptimal experience
• Experience training or post-training VLAs, WAMs, or other foundation models for embodied decision-making
• Experience with diffusion or flow-based policies, action-sequence prediction, model-based RL, or action-conditioned world models
• Hands-on experience with distributed training across multiple GPUs or nodes, including improving training throughput, memory efficiency, and stability
• Experience evaluating learned policies on physical robots or robotics simulation benchmarks
• Proven track record of achieving significant results as demonstrated by grants, fellowships, patents, and/or accepted papers in top conferences or journals such as NeurIPS, ICML, ICLR, CoRL, RSS, ICRA, CVPR, ICCV, ECCV, ACL, etc