PeopleofSuperintelligence

Researcher

Sanjiban Choudhury

Sanjiban Choudhury is a Cornell computer scientist whose research helps robots and AI agents learn from people. At Carnegie Mellon, he studies robot planning, becomes a Siebel Scholar and completes a Ph.D. at the Robotics Institute. His research spans imitation learning, the RHyME robot-learning system and LEAP, a framework that trains language-model agents with AI feedback. As of October 2026, he is an assistant professor of computer science at Cornell and leads the PoRTaL group.

Works
101
Citations
1,368
h-index
19

Publication numbers from OpenAlex, as of 27 October 2026.

Timeline

  1. 2013

    Robotics and a Siebel Scholar

    In September 2012, Carnegie Mellon names him to its 2013 class of Siebel Scholars. A graduate of IIT Kharagpur, he studies motion planning and relative pose estimation for autonomous helicopters in the Robotics Institute's master's program.

  2. May 2018

    Adaptive Motion Planning

    He completes a Ph.D. at Carnegie Mellon's Robotics Institute in May 2018 with the thesis Adaptive Motion Planning, advised by Sebastian Scherer.

  3. March 2022

    Aurora and Cornell roles

    In March 2022, he is a research scientist at Aurora Innovation and a soon-to-be assistant professor at Cornell University. His career includes a postdoctoral fellowship at the University of Washington, and his research focuses on imitation learning, decision making and human-robot interaction.

  4. September 2023

    Superhuman imitation learning

    In September 2023, he co-leads a project that aims to train robots from imperfect demonstrations and improve on them. The National Science Foundation supports the three-year project with nearly $1.2 million.

  5. April 2024

    OpenAI superalignment grant

    In April 2024, OpenAI awards him $150,000 to study task superalignment. He plans to combine imitation learning with language models that generate code, testing whether supervision on easy Python tasks can extend to complex C++ programs.

  6. October 2024

    LEAP trains LLM agents

    In October 2024, he and Paloma Sodhi introduce LEAP, a fine-tuning framework that trains language-model agents with feedback from AI teachers who have privileged information during training. Experiments show that smaller student models can outperform stronger teachers on text games, web navigation and coding tasks.

  7. 2025

    An ONR robotics award

    The 2025 ONR Young Investigator Program names him among its 24 recipients and funds his work to train robots for multistep tasks. His project combines repair manuals, videos of people and real-time feedback for jobs such as ship-engine inspection and repair.

  8. April 2025

    RHyME learns from video

    In April 2025, he is the senior author of Cornell's RHyME work, which lets a robot learn tasks from a single how-to video by matching the demonstration with related examples in stored videos. In lab tests, RHyME uses 30 minutes of robot data and raises task success by more than 50% over previous methods.

  9. 2026

    An NSF CAREER award

    In January 2026, he receives an NSF Faculty Early Career Development award for robots that learn new skills by watching people, trying tasks and improving from feedback. The project includes K-12 robotics activities and online resources for STEM participation.

  10. As of October 2026

    Professor at Cornell

    As of October 2026, he is an assistant professor of computer science at Cornell and leads the PoRTaL group. His research covers interactive AI agents, reinforcement learning from feedback, imitation learning and foundation models for planning, robotics and code generation.

Career

  1. NowCornell UniversityAssistant Professor of Computer Science, Group Lead
  2. University of WashingtonPostdoctoral Fellow
  3. Aurora InnovationResearch Scientist

Education

  1. 2018Carnegie Mellon UniversityPh.D.
  2. Indian Institute of Technology Kharagpur
  3. Carnegie Mellon UniversityMaster's

From the biography.

Sources and credits

Photo: We have no published photo of Sanjiban Choudhury yet. Send us a free-licensed photo.

  1. Five SCS Students Named Siebel Scholars cs.cmu.edu. Checked 26 October 2026.
  2. Sanjiban Choudhury - Robotics Institute Carnegie Mellon University ri.cmu.edu. Checked 26 October 2026.
  3. Interactive Imitation Learning: Planning Alongside Humans robotics.cornell.edu. Checked 26 October 2026.
  4. When robots imitate life: Project explores better way to train AI news.cornell.edu. Checked 26 October 2026.
  5. OpenAI funds Choudhury's work to keep superhuman AI in check bowers.cornell.edu. Checked 26 October 2026.
  6. Better than Your Teacher: LLM Agents that learn from Privileged AI Feedback arxiv.org. Checked 26 October 2026.
  7. Choudhury wins Navy Young Investigator award to train robots news.cornell.edu. Checked 26 October 2026.
  8. Robot see, robot do: System learns after watching how-tos news.cornell.edu. Checked 26 October 2026.
  9. Three early-career professors win NSF development awards news.cornell.edu. Checked 26 October 2026.
  10. Sanjiban Choudhury | Cornell Bowers bowers.cornell.edu. Checked 26 October 2026.