PeopleofSuperintelligence

Researcher

Kuo-Hao Zeng

Also known as 曾國豪

Kuo-Hao Zeng is a computer vision and embodied AI researcher whose work trains robots and simulated agents to forecast, navigate and handle objects; since February 2026 he has been a member of the technical staff at Anthropic. After engineering studies at National Sun Yat-sen University and National Tsing Hua University, he completed a PhD in 2023 at the University of Washington's RAIVN Lab with Ali Farhadi and Roozbeh Mottaghi, where one of his projects taught a simulated drone to play catch. At the Allen Institute for AI from 2023 to 2025, he was first author of PoliFormer, a navigation agent that won an Outstanding Paper Award at CoRL 2024, and he coauthored the open Molmo vision-language models. In March 2025, he became a founding research scientist at Vercept, where he worked on computer use.

Now at
Anthropic
Works
30
Citations
1,029
h-index
15

Publication numbers from OpenAlex, as of 22 October 2026.

Timeline

  1. 2010–2017

    Studies at Sun Yat-sen and Tsing Hua

    He studies mechanical and electromechanical engineering at National Sun Yat-sen University from 2010 to 2014. From 2014 to 2017, he is an electrical engineering master's student at National Tsing Hua University, advised by Min Sun in the Vision Science Lab.

  2. 2016–2017

    Video titles and accident anticipation

    With Tseng-Hung Chen, Juan Carlos Niebles and Min Sun, he presents automatic title generation for user-generated videos at ECCV 2016 and collects the Video Titles in the Wild dataset of 18,100 videos and titles. At CVPR 2017, he is first author of a paper on accident anticipation that introduces the Epic Fail dataset of 3,000 viral videos.

  3. 2018–2020

    Teaching a drone to catch

    He begins a computer science PhD at the University of Washington in 2018, advised by Ali Farhadi and Roozbeh Mottaghi in the RAIVN Lab. At CVPR 2020, his visual reaction paper teaches a drone to play catch in simulated environments, using a model that combines a forecaster with a planner and a new dataset of 30,000 throws of 20 object types.

  4. 2021

    A J.P. Morgan PhD fellowship

    In 2021, he is named a J.P. Morgan Ph.D. Fellow for his research on forecasting for decision making in AI. His CVPR 2021 paper on interactive visual navigation trains agents with reinforcement learning to unblock paths toward a target, using a neural physics model that learns the outcome of actions.

  5. 2023–2025

    Research scientist at Ai2

    He completes his PhD in 2023 and works as a research scientist on embodied AI at the Allen Institute for AI (Ai2) from 2023 to 2025. He is one of six equal-contribution first authors of SPOC, a CVPR 2024 paper whose robot agent learns to navigate, explore and manipulate objects by imitating shortest paths in simulation.

  6. 2024

    PoliFormer wins at CoRL

    He is first author of PoliFormer, an RGB-only indoor navigation agent trained with reinforcement learning at scale purely in simulation that generalizes to the real world without adaptation. PoliFormer wins an Outstanding Paper Award at CoRL 2024, and Open X-Embodiment, a robot-learning dataset collaboration he coauthors, wins the ICRA 2024 Best Paper Award.

  7. 2024–2025

    Coauthoring Molmo

    He coauthors Molmo, a family of open vision-language models first posted in September 2024 together with PixMo, a collection of new datasets gathered without the use of external vision-language models. The Molmo and PixMo paper receives a Best Paper Honorable Mention at CVPR 2025.

  8. Since February 2026

    Technical staff at Anthropic

    From March 2025 to February 2026, he is a founding research scientist at Vercept, where he works on computer use. In February 2026, he joins Anthropic as a member of the technical staff, working on computer vision, embodied AI and computer-use agents.

Career

  1. Since 2026AnthropicMember of the technical staff
  2. 2025–2026VerceptFounding Research Scientist
  3. 2023–2025Allen Institute for AI (Ai2)Research Scientist

Education

  1. From 2018University of WashingtonPhD
  2. 2017National Tsing Hua UniversityMaster's
  3. 2014National Sun Yat-sen University

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. Open X-Embodiment: Robotic Learning Datasets and RT-X Models: Open X-Embodiment Collaboration 02024 · IEEE International Conference on Robotics and Automation · 187 citationsWith Animesh Garg, Aniruddha Kembhavi, Bernhard Schölkopf, Cheng Chi, Deepak Pathak and Gaurav S. Sukhatme
  2. Leveraging Video Descriptions to Learn Video Question Answering2017 · 162 citations
  3. Open X-Embodiment: Robotic Learning Datasets and RT-X Models2023 · 110 citationsWith Animesh Garg, Bernhard Schölkopf, Deepak Pathak, Gaurav S. Sukhatme, Henrik I. Christensen and Ilija Radosavovic
  4. Omnidirectional CNN for Visual Place Recognition and Navigation2018 · IEEE International Conference on Robotics and Automation · 80 citations
  5. Title Generation for User Generated Videos2016 · Lecture notes in computer science · 78 citations

From OpenAlex, as of 21 October 2026.

  1. The median h-index is 48 at universities and 14 at frontier AI labsIncludes Kuo-Hao Zeng. The median h-index is 48 at universities and 14 at frontier AI labs. We compare publishing careers, recent citations, dual appointments and job moves across 2,420 university researchers and 308 frontier-lab researchers.
  2. 22 of 133 people shaping AI with a former OpenAI role now work at AnthropicIncludes Anthropic. Our estimates of where OpenAI recruits came from and where 133 people with former OpenAI roles work now, including the companies and labs they founded.
  3. Companies supply the most prior-work hires at 7 of the 9 frontier AI labsIncludes Anthropic. Our estimates of the organizations where 550 people worked directly before joining 9 frontier AI labs across 620 career transitions, comparing the main feeder organizations per lab.
  4. Of 943 people shaping AI who worked at Google, 446 work there nowIncludes Anthropic. Our estimates of where the 943 people shaping AI who worked at Google and Google DeepMind work now: 446 at Google and 452 across 278 other organizations.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

Photo: We have no published photo of Kuo-Hao Zeng yet. Send us a free-licensed photo.

  1. Kuo-Hao Zeng | alphaXiv alphaxiv.org. Checked 20 October 2026.
  2. [1608.07068] Title Generation for User Generated Videos arxiv.org. Checked 20 October 2026.
  3. Agent-Centric Risk Assessment: Accident Anticipation and Risky Region Localization aliensunmin.github.io. Checked 20 October 2026.
  4. Visual Reaction: Learning to Play Catch With Your Drone (CVPR 2020 Open Access Repository) openaccess.thecvf.com. Checked 20 October 2026.
  5. Allen School student Kuo-Hao Zeng named J.P. Morgan Ph.D. Fellow for his work in visual forecasting in artificial intelligence news.cs.washington.edu. Checked 20 October 2026.
  6. SPOC: Imitating Shortest Paths in Simulation Enables Effective Navigation and Manipulation in the Real World spoc-robot.github.io. Checked 20 October 2026.
  7. Kuo-Hao Zeng kuohaozeng.github.io. Checked 20 October 2026.
  8. PoliFormer: Scaling On-Policy RL with Transformers Results in Masterful Navigators proceedings.mlr.press. Checked 20 October 2026.
  9. CoRL 2024 - Awards 2024.corl.org. Checked 20 October 2026.
  10. [2409.17146] Molmo and PixMo: Open Weights and Open Data for State-of-the-Art Vision-Language Models arxiv.org. Checked 20 October 2026.