Robotics Intern, Agentic Robotics

Toyota Research Institute•Cambridge, MA
•$40 - $58•Hybrid

About The Position

Toyota Research Institute is developing general-purpose robots capable of performing a wide variety of dexterous tasks. The Large Behavior Models team builds machine-learning foundation models for dexterous robot manipulation, using generative AI to produce robot actions from sensor data and human requests. Its work combines embodied robot demonstrations with large text, image, and video datasets, and uses high-quality simulation to supplement real-world data with procedurally generated synthetic demonstrations. The research intern will work on data-efficient, general algorithms for learning robust policies from multiple sensing modalities, including proprioception, images, force, audio, and dense tactile sensing. The role involves developing working code and/or hardware prototypes, collaborating frequently with team members, running experiments with simulated and physical robots, and contributing to peer-reviewed publications. The intern should be comfortable working with both existing large static datasets and a growing, dynamic corpus of robot data. This is a paid, 12-week hybrid internship based in an in-office environment. The expected pay at commencement is $40–$58 per hour for Massachusetts-based roles. TRI offers medical, dental, and vision insurance, as well as paid time off including holiday pay and sick time.

Requirements

  • Research experience in classical robotics, robot learning, reinforcement learning, world models, or embodied reasoning.
  • Interest in connecting reasoning and physical action.
  • Strong software development skills in Python.
  • Experience with modern deep learning frameworks such as PyTorch.
  • Comfort with rapid prototyping and testing in simulation and on physical robots.

Nice To Haves

  • Experience with large-model post-training, simulation-based reinforcement learning, sim-to-real transfer, or contact-rich manipulation.
  • A creative approach to training and adapting robotics foundation models, including reinforcement learning in simulation, action-conditioned prediction, and transfer to new tasks and physical robots.

Responsibilities

  • Research data-efficient, general algorithms for learning robust robot policies using multiple sensing modalities.
  • Create working code and/or hardware prototypes.
  • Collaborate frequently with team members.
  • Run experiments with simulated and physical robots.
  • Participate in publishing work in peer-reviewed venues.
  • Work with large static datasets and a growing, dynamic corpus of robot data.

Benefits

  • Medical insurance
  • Dental insurance
  • Vision insurance
  • Paid time off, including holiday pay and sick time
© 2026 Teal Labs, Inc
Privacy PolicyTerms of Service