Postdoctoral Researcher, Post-Training and Continual Adaptation for Robotics

At Toyota Research Institute (TRI), we’re on a mission to improve the quality of human life. We’re developing new tools and capabilities to amplify the human experience. To lead this transformative shift in mobility, we’ve built a world-class team advancing the state of the art in AI, robotics, driving, and material sciences.

We are seeking a postdoctoral researcher to advance post-training and adaptation for generative robot policies. Our goal is to enable pretrained policies to improve task performance, adapt to human preferences, and acquire new behaviors from limited experience while preserving existing capabilities.

You will work closely with our robotics and machine learning researchers on a focused project, developing new methods and validating them through experiments in simulation and on real robots.

Pretrained robot policies can perform increasingly broad behaviors, yet improving them for new tasks, environments, and user needs remains difficult. Robot experience is limited, feedback can be imperfect, and adapting a policy can undermine capabilities it already has.

Research directions include learning efficiently from demonstrations, online interaction, and human feedback; using reinforcement learning and related methods to improve and customize pretrained policies; and retaining useful improvements across tasks and over time. The project will pursue a specific question within this agenda, shaped jointly around the researcher’s expertise and the team’s research priorities.

Responsibilities

Develop methods for post-training and adapting pretrained robot policies, taking a focused research question from initial formulation through experimental validation.

Design rigorous experiments to evaluate policy improvement, behavioral adaptation, robustness, and retention of existing capabilities relative to robot experience and human effort.

Collaborate with the research team and communicate findings through publications in leading robotics and machine learning venues.

Qualifications

Ph.D. in robotics, machine learning, computer science, or a related field, completed by the start of the appointment.

Strong research background in reinforcement learning, imitation learning, or related robot-learning methods, demonstrated through relevant publications.

Strong programming and experimental skills, including experience with deep learning frameworks such as PyTorch or JAX and with robotic systems.

Ability to develop and carry out a scientific research plan in collaboration with other researchers.

Experience with robot foundation models, online adaptation, human feedback, or continual learning is especially relevant.