Reinforcement Learning Intern

Intuitive•Sunnyvale, CA

About The Position

The Reinforcement Learning Intern will report to the future forward research lead and contribute to developing and evaluating reinforcement learning (RL) algorithms in support of supervised autonomy initiatives. The intern will identify, train, and test RL algorithms using data collected from Intuitive surgical robot systems. Most work will use offline-collected data rather than real-time robot operation, with real-time deployment as a stretch goal. The intern will collect trajectory data through tele-operated simulation and/or real surgical robot exercises, train and benchmark established RL algorithms, adapt algorithms to the surgical domain, and recommend an algorithm and experimental setup based on results. The intern may also deploy a selected algorithm in simulation or on the dV5 to validate real-time performance and will present findings to the broader team. The internship requires full-time availability, approximately 40 hours per week, for 10–12 weeks starting in May or June. Applicants must be enrolled in and returning to an accredited degree-seeking academic program after the internship. The required program of study is at the doctorate level in engineering, robotics, computer science, or a related field.

Requirements

Must be currently enrolled in and returning to an accredited degree-seeking academic program after the internship.
Must be available to work full-time, approximately 40 hours per week, for 10–12 weeks starting in May or June; specific start dates are shared during recruiting.
Currently enrolled in a doctorate-level degree-seeking program in engineering (such as biomedical or electrical engineering), robotics, computer science, or a related field.

Nice To Haves

Deep experience with a deep learning framework such as PyTorch or JAX.
Strong familiarity with reinforcement learning algorithms, such as policy gradient or actor-critic methods.
Willingness to learn about robotics and applying reinforcement learning to surgical robotics.
Strong written and verbal communication skills for explaining experiment motivations, findings, and implications.
Experience with video-based and multimodal data analysis and processing.

Responsibilities

Collect tele-operated simulation and/or real data from the surgical robot while performing a realistic exercise, such as a Fundamentals of Laparoscopic Surgery pick-and-place exercise.
Train off-the-shelf reinforcement learning algorithms on trajectory data and benchmark their performance using established metrics.
Adapt reinforcement learning algorithms to the surgical domain and exercise of interest, and identify techniques to improve performance relative to off-the-shelf models.
Recommend an algorithm and experimental setup for future work based on experimental results, and present findings to the broader team.
Deploy a selected RL algorithm in simulation or on the dV5 to validate its performance in real time, if feasible.