Research Scientist Intern, Robotics Foundation Models (PhD)

Meta Reality Labs Research is a team of researchers pushing the frontier of AI and robotics. Our team is building the next generation of foundation models for robotic manipulation, including vision-language-action models (VLAs) and world action models (WAMs). We are seeking a Research Scientist Intern to help design and improve the architectures of these models.

Interns in this role may work across policy modeling and architecture: action representations, diffusion and flow-based action heads, integration with pretrained vision-language and video backbones, and efficient inference for real-time control. It involves prototyping new architectures together with research scientists and engineers, running the experiments that show which design choices matter, and testing the most promising models in simulation and on real robots.

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

Responsibilities

Design and evaluate policy architectures for WAMs and VLAs, including action representations, diffusion and flow-based action heads, and integration with pretrained vision-language and video backbones. Study how architecture choices affect generalization across tasks and embodiments, scaling with data and compute, and inference latency. Develop techniques that make large policies fast enough for real-time control on robots. Train models at scale on large, heterogeneous robot and video datasets, including multi-node distributed training. Design controlled experiments and ablations that isolate which design choices matter, and evaluate models in simulation and on real robots. Collaborate with research scientists and engineers to bring the most promising architectures into the team’s foundation models. Communicate research findings and publish results at leading machine learning, computer vision, and robotics venues.

Qualifications

Currently has or is in the process of obtaining a Ph.D. degree in Computer Science, Computer Engineering, Artificial Intelligence, Robotics, Computer Vision, Machine Learning or related STEM field Experience with any of the following research areas: robotics, motion planning, embodied AI, human-robot interaction, sim-to-real transfer, learning from demonstration, reinforcement learning, dexterous manipulation, digital agents, vision language models, computer vision, egocentric perception, and/or LLMs Experience with machine learning models, and frameworks such as PyTorch, TensorFlow, etc 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 designing and training policy architectures for robot learning, such as vision-language-action models, world action models, or diffusion and flow-based policies Experience with architecture design or generative modeling in adjacent domains, such as video generation, vision-language models, or LLMs Hands-on experience with distributed training across multiple GPUs or nodes, including improving training throughput, memory efficiency, and stability 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), Robotics (ICRA, IROS, RSS, CoRL), Computer Vision (CVPR, ICCV, ECCV) Experience in robotics or related field, or advanced degree with equivalent experience Experience with manipulating and analyzing complex, large scale, high-dimensionality data from varying sources Experience solving complex problems and comparing alternative solutions, trade-offs, and diverse points of view to determine a path forward Experience working and communicating cross functionally in a team environment Experience building systems based on machine learning and/or deep learning methods