PeopleofSuperintelligence

Researcher

Ching-An Cheng

Ching-An Cheng is a reinforcement-learning researcher whose work connects robotics, offline reinforcement learning, and agents that learn from feedback. He earned bachelor's degrees in mechanical and electrical engineering at National Taiwan University in 2011, an M.S. in mechanical engineering there in 2013, and a robotics PhD at Georgia Tech in 2020. His robot-learning work won the AISTATS Best Paper Award in 2018, and he shared a NeurIPS workshop paper award in 2019. He coauthored two offline reinforcement-learning papers in 2021 and 2022, including ATAC, which received an ICML Outstanding Paper Runner Up award; in 2024 he helped release Trace, an open-source Python library for training AI systems with feedback. In July 2025, he co-organized an ICML agent-learning workshop; as of October 2026, he is a Senior Research Scientist in Google Research's Learning Theory Group and coauthor of POLCA.

Badges

Timeline

  1. 2011–2013

    Degrees at National Taiwan University

    He earns bachelor's degrees in mechanical and electrical engineering at National Taiwan University in 2011 and an M.S. in mechanical engineering there in 2013.

  2. 2018–2020

    A Georgia Tech robotics PhD

    At Georgia Tech's Institute for Robotics and Intelligent Machines, he studies efficient robot learning as a robotics PhD student advised by Byron Boots. During his PhD, he interns at Nvidia Research in 2018 and Microsoft Research AI in 2019, then completes the robotics doctorate at Georgia Tech in 2020. His 2018 paper with Boots, “Convergence of Value Aggregation for Imitation Learning,” wins the AISTATS Best Paper Award.

  3. 2019

    A NeurIPS workshop paper award

    He shares the 2019 Best Paper Award at the NeurIPS Optimization Foundations for Reinforcement Learning Workshop with Jonathan Lee, Ken Goldberg and Byron Boots for “Continuous Online Learning and New Insights to Online Imitation Learning.”

  4. 2021

    Bellman-consistent offline RL

    In 2021, he coauthors “Bellman-consistent Pessimism for Offline Reinforcement Learning” with Tengyang Xie, Nan Jiang, Paul Mineiro and Alekh Agarwal. The paper introduces Bellman-consistent pessimism for general function approximation and applies it at the initial state over functions consistent with the Bellman equations.

  5. 2022

    Offline RL with ATAC

    With Tengyang Xie, Nan Jiang and Alekh Agarwal, he coauthors “Adversarially Trained Actor Critic for Offline Reinforcement Learning” in 2022. The paper proposes ATAC, a model-free offline RL algorithm framed as a two-player Stackelberg game, and receives an ICML Outstanding Paper Runner Up award.

  6. 2024

    Trace for AI systems

    In July 2024, Cheng, Allen Nie and Adith Swaminathan release Trace, an open-source Python library and generative optimization framework for training AI systems with feedback. Trace uses numerical losses, natural language and execution errors; the team demonstrates it with a simulated robot controller.

  7. July 2025

    An ICML agent-learning workshop

    At ICML in July 2025, he co-organizes the Programmatic Representations for Agent Learning workshop as a Google Research organizer.

  8. As of October 2026

    Google Research scientist

    As of October 2026, he is a Senior Research Scientist in Google Research's Learning Theory Group, after serving as a Principal Researcher in Microsoft Research AI Frontiers. In 2026, he coauthors POLCA, a stochastic generative optimization method that uses a language model with numerical rewards and text feedback to optimize complex systems.

Career

  1. NowGoogle ResearchSenior Research Scientist
  2. 2018NVIDIAResearch Intern
  3. Microsoft ResearchResearch Intern

Education

  1. 2020Georgia Institute of TechnologyRobotics PhD
  2. 2013National Taiwan UniversityM.S. in mechanical engineering
  3. 2011National Taiwan UniversityBachelor's degree in mechanical engineering

From the biography.

  1. 16% of AI career moves in the 2020s went from a university to a companyIncludes Ching-An Cheng. 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. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other universityIncludes Ching-An Cheng. The Massachusetts Institute of Technology trained or employed 44 of the 187 AI major-prize winners, more than any other university. We compare doctorate and 5+ year employment links, career timing and current workplaces for 187 major-prize and 1,111 best-paper winners.
  3. Google had the most people shaping AI every year from 2014 to 2025Includes Google. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  4. Google has the highest combined h-index among AI organizationsIncludes Google. 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.

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 Ching-An Cheng yet. Send us a free-licensed photo.

  1. Ching-An Cheng chinganc.com. Checked 22 October 2026.
  2. Policy Optimization as Predictable Online Learning Problems: Imitation Learning and Beyond microsoft.com. Checked 22 October 2026.
  3. NeurIPS 2019 Optimization Foundations for Reinforcement Learning Workshop optrl2019.github.io. Checked 22 October 2026.
  4. Bellman-consistent Pessimism for Offline Reinforcement Learning papers.neurips.cc. Checked 22 October 2026.
  5. ICML 2022 Awards icml.cc. Checked 22 October 2026.
  6. Tracing the path to self-adapting AI agents microsoft.com. Checked 22 October 2026.
  7. Programmatic Representations for Agent Learning Workshop at ICML 2025 pral-workshop.github.io. Checked 22 October 2026.
  8. POLCA: Stochastic Generative Optimization with LLM research.google. Checked 22 October 2026.