PeopleofSuperintelligence

Researcher

Jiefeng Li

Also known as 李杰锋

Jiefeng Li is a computer vision researcher who builds models of human pose and motion, and he is a senior research scientist in the Data-Driven AI for Robotics (DAIR) team at NVIDIA Research. At Shanghai Jiao Tong University, he earned a bachelor's degree in electronic engineering in 2018 and a master's in computer science in 2020, then completed a PhD in 2024 in Cewu Lu's MVIG lab. His SJTU work included the CrowdPose benchmark for crowded scenes in 2019 and HybrIK, a 2021 inverse-kinematics method for recovering 3D body meshes, and he helped maintain the AlphaPose pose-tracking system. In May 2025, he posted GENMO, a single model that both estimates and generates human motion, and it appeared at ICCV in October 2025. He also coauthored SONIC, a humanoid whole-body controller that Science Robotics published in July 2026.

Now at
NVIDIA
Badges

Timeline

  1. 2014–2020

    Two degrees at SJTU

    He studies electronic engineering at Shanghai Jiao Tong University from 2014 and earns his bachelor's degree in 2018. His master's degree in computer science, also from SJTU, follows in 2020.

  2. 2018–2019

    Pose Flow and CrowdPose

    He coauthors Pose Flow, an efficient online pose-tracking method, with Yuliang Xiu and Cewu Lu at BMVC 2018. At CVPR 2019, he is first author of CrowdPose, a pose-estimation method and benchmark for crowded scenes.

  3. 2021

    HybrIK and residual log-likelihood estimation

    At CVPR 2021, he presents HybrIK, a hybrid analytical-neural inverse kinematics method that turns 3D joints into body-part rotations to rebuild a 3D body mesh. His Residual Log-likelihood Estimation paper, an ICCV 2021 oral, lets a regression method outperform heatmap-based methods on multi-person pose estimation for the first time. In September 2021, he gives a talk at the Max Planck Institute for Intelligent Systems.

  4. 2022

    The AlphaPose system

    He coauthors the 2022 IEEE Transactions on Pattern Analysis and Machine Intelligence paper on AlphaPose, the MVIG-SJTU system for real-time whole-body multi-person pose estimation and tracking. He is also one of the system's maintainers.

  5. 2024

    A PhD in the MVIG lab

    He receives his PhD in 2024 from the Department of Computer Science at SJTU, where he is a member of the MVIG lab supervised by Cewu Lu.

  6. 2025

    GENMO, a generalist motion model

    As a research scientist in NVIDIA Research's DAIR team, he is first author of GENMO, posted in May 2025, which unifies human motion estimation and generation in one model. The paper appears at the IEEE International Conference on Computer Vision in October 2025.

  7. 2026

    SONIC in Science Robotics

    In March 2026, he coauthors two DAIR preprints: Kimodo, on scaling controllable human motion generation, and SOMA, which unifies parametric human body models. SONIC, a foundation model for humanoid whole-body control trained on 700 hours of motion capture, appears in Science Robotics in July 2026 with him as a coauthor.

  8. As of October 2026

    Senior research scientist at NVIDIA

    He is a senior research scientist in the Data-Driven AI for Robotics (DAIR) team at NVIDIA Research. His research covers human motion, human-object interaction, humanoids and embodied AI.

Career

  1. Since 2026NVIDIAResearch Scientist
  2. Until 2024Shanghai Jiao Tong UniversityMember

Education

  1. 2024Shanghai Jiao Tong UniversityPhD
  2. 2020Shanghai Jiao Tong UniversityMaster's degree
  3. 2018Shanghai Jiao Tong UniversityBachelor's degree

From the biography.

  1. 16% of AI career moves in the 2020s went from a university to a companyIncludes Jiefeng Li. We estimate that 16% of AI career moves in the 2020s went from a university to a company; 29% of moves starting at a university went to a company, and 76% ended the university role.
  2. Google has the highest combined h-index among AI organizationsIncludes NVIDIA. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.
  3. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes NVIDIA. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.
  4. Google alumni have founded or led 133 other organizations, more than any other AI labIncludes NVIDIA. We estimate that alumni of 17 major AI labs went on to found or lead 350 other organizations, including 226 they founded, and we trace 472 career moves directly between 23 labs since 2010.

Colleagues

People with dated positions at the same organization in the same years.

Sources and credits

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

  1. Jiefeng Li | Research research.nvidia.com. Checked 20 October 2026.
  2. Jiefeng Li | DAIR @ NVIDIA Research research.nvidia.com. Checked 20 October 2026.
  3. GitHub - MVIG-SJTU/AlphaPose: Real-Time and Accurate Full-Body Multi-Person Pose Estimation&Tracking System github.com. Checked 20 October 2026.
  4. CVPR 2019 Open Access Repository: CrowdPose: Efficient Crowded Scenes Pose Estimation and a New Benchmark openaccess.thecvf.com. Checked 20 October 2026.
  5. [2011.14672] HybrIK: A Hybrid Analytical-Neural Inverse Kinematics Solution for 3D Human Pose and Shape Estimation arxiv.org. Checked 20 October 2026.
  6. [2107.11291] Human Pose Regression with Residual Log-likelihood Estimation arxiv.org. Checked 20 October 2026.
  7. Jiefeng Li jeffli.site. Checked 20 October 2026.
  8. [2211.03375] AlphaPose: Whole-Body Regional Multi-Person Pose Estimation and Tracking in Real-Time arxiv.org. Checked 20 October 2026.
  9. [2505.01425] GENMO: A GENeralist Model for Human MOtion arxiv.org. Checked 20 October 2026.
  10. GENMO: A GENeralist Model for Human MOtion | DAIR @ NVIDIA Research research.nvidia.com. Checked 20 October 2026.
  11. Kimodo: Scaling Controllable Human Motion Generation | DAIR @ NVIDIA Research research.nvidia.com. Checked 20 October 2026.
  12. SOMA: Unifying Parametric Human Body Models | DAIR @ NVIDIA Research research.nvidia.com. Checked 20 October 2026.
  13. SONIC: Supersizing Motion Tracking for Natural Humanoid Whole-Body Control | DAIR @ NVIDIA Research research.nvidia.com. Checked 20 October 2026.
  14. [2511.07820] SONIC: Supersizing Motion Tracking for Natural Humanoid Whole-Body Control arxiv.org. Checked 20 October 2026.