I am a postdoc at Stanford University advised by Chelsea Finn. I am broadly interested in building general-purpose robots that can perform any tasks in the real-world, ranging from surgical to humanoid applications.
Previously, I focused on developing dextrous manipulation policies for the da Vinci robot with Axel Krieger. I also worked on developing autonomous surgical workflows for eye surgery with Marin Kobilarov and Iulian Iordachita. I interned at Zoox (self-driving company owned by Amazon) in the planning & controls and prediction team.
jwbkim@stanford.edu
UpcomingGiving a talk at the Georgia Chapter of the American College of Cardiology.
UpcomingI will be a panelist at the IROS 2026 Evolving Landscape of Surgical Robotics Workshop.
Our paper Ego-Pi is released! We explore how we can teach robots new tasks using human data.
Giving another talk at Samsung Medical Center in Seoul, Korea.
Giving a talk at the Yale AI in Surgical Care Symposium.
Giving a guest lecture in the Johns Hopkins University robot learning course!
Giving a talk at the UPenn GRASP seminar.
Giving a talk at the Global Breast Cancer Conference (GBCC) in Seoul, Korea.
Giving a talk at Samsung Medical Center in Seoul, Korea.
Giving a talk at the SNU AI Computing School in Seoul, Korea.
Giving a talk at NeurIPS GenAI4Health in San Diego, CA.
Giving a talk at the Clinical Robotic Surgical Association (CRSA).
Giving a talk at CoRL’s Automating Robotic Surgery Workshop in Seoul, Korea.
Giving a talk at the China GI Surgical Forum.
Our Science Robotics paper SRT-H made the July cover and ranked 2nd in Altmetric score among all Science Robotics papers. More than 280 news articles have been written about the work!
Giving a talk at the Korean Society of Endo-Laparoscopic & Robotic Surgery (KSERS) in Seoul, Korea.
Our Surgical Robot Transformer (SRT) paper has been covered in the Washington Post, among many other news outlets!
Giving an oral presentation at the Conference on Robot Learning on Surgical Robot Transformer (SRT) in Munich, Germany.
Scroll for earlier updates
CVPR Findings · 2026
We show that we can teach robots new tasks, such as new sorting rules, skill composition, and rule-based ordering, via co-training with human data. This is useful because if you want to steer your robot to perform new tasks, you can simply collect human data rather than further robot data.
Science Robotics · 2025
July 2025 cover
2nd highest Altmetric score among Science Robotics papers (July 2025)
We explore language-conditioned hierarchical imitation learning for real surgery using animal tissues.
CoRL · 2024
Oral presentation · 4.5% acceptance rate
We explore whether surgical manipulation tasks can be learned on the da Vinci system via imitation learning. IL on da Vinci turns out to be non-trivial, due to forward kinematics errors that can reach +/- 5cm.
27 publications
No publications match this search.