AI Research Scientist, Embodied AI (PhD)

Meta is seeking a researcher to join Meta Fundamental AI Research (FAIR), a research organization focused on making significant progress in AI. Individuals in this role are expected to be recognized experts in identified research areas such as artificial intelligence, machine learning, robotics, and embodied AI, particularly including areas such as transfer learning, learning from demonstration, reinforcement learning, perception, representation learning, robot control, navigation, mobile manipulation, dexterous manipulation, vision-language models, and contextually aware AI. The ideal candidate should have a keen interest in producing new, open science to make embodied agents more intelligent. Some research directions of interest are: - Investigate algorithmic development and application of foundation models for embodied AI and/or robotics - Representation learning for Embodied AI tasks - Developing the next generation of simulation platforms for embodied AI research (e.g. AI Habitat) - Sim-to-real transfer - Planning and learning for dynamic in-hand manipulation tasks - Performing broad dexterous and functional manipulation (rigid to deformable objects; in-hand to against the environment) in a generalizable manner - Tactile and multimodal perception - Continual learning in interactive environments - Long-horizon reasoning for Embodied AI tasks in human environments given natural-language instructions, like “clean up the house”, or “where are my glasses” - Developing efficient models deployable to robots and hardware