Quanming Yao, 姚权铭

Quanming Yao (姚权铭)

Associate Professor & Ph.D. Advisor

Department of Electronic Engineering, Tsinghua University

Also affiliated as a Ph.D. supervisor with the Zhongguancun Institute of Artificial Intelligence

Also affiliated with the State Key Laboratory of Space Network and Communications.

Also affiliated with the Beijing National Research Center for Information Science and Technology.

E-mail: qyaoaa [AT] connect.ust.hk / tsinghua.edu.cn

Office: 11-305 Room, Rohm Building, Tsinghua. Beijing, China, 100084 (MAP)

Group Code Repository on GitHub, RedNote (小红书)

About Me

Dr. Quanming Yao is currently an associate professor in the Department of Electronic Engineering, Tsinghua University. Previously, he was a senior scientist at 4Paradigm Inc., where he founded and led its machine learning research team. He earned his Ph.D. from the Hong Kong University of Science and Technology (HKUST) and his bachelor's degree from Huazhong University of Science and Technology (HUST).

His honors include IEEE's Top 30 Early Career to Watch, Ant Group's inaugural Intech Prize, the INNS Aharon Katzir Young Investigator Award, Forbes 30 Under 30 China, and a Google Fellowship. He is an IET Fellow, serves as Program Co-chair of ADMA 2026, and is an Action Editor for IEEE Transactions on Pattern Analysis and Machine Intelligence and a Senior Action Editor for Neural Networks. He has also been recognized with the EE Department's Zheng Junli Outstanding Teacher Award and Tsinghua University's Liu Bing Award.

Research Focus

Our research lies in core machine learning, especially scalable and data-efficient learning. I develop principled algorithms for learning under limited supervision, limited computation, and structured data, with applications in recommendation, knowledge graphs, scientific discovery, and large-model systems.

To get a sense of how we work, see this paper. To learn what we work on, see this survey.
Recent talks on LLM Training Dynamics and Agentic Few-shot Learning.

Current Research Topics: Data-Efficient Agentic Learning · Few-Shot and In-Context Learning · LLM Training Dynamics

Recruitment

Various positions (postdoctoral researchers / Ph.D. and master's students / research engineers / research assistants) are available. You can join us through:

Electrical Engineering Department of Tsinghua University
Qiuzhen College, Tsinghua University (with Dr. Yaqing Wang)
Zhongguancun Institute of Artificial Intelligence (select me as your research mentor)

You may read this guide to prepare to work with our group.

Recent News --- old ones ---

2026.08: Gave a talk on "Understanding Low-Precision LLM Training Stability" at "CSML".
2026.08: Gave a talk on "Data-Efficient Learning in the Era of Agents" at the "CSIG Young Scientists Conference".
2026.07: Will serve as an Area Chair for AAAI 2027.
2026.07: Received the Third Prize in the Electronic Information Teaching Competition.
2026.06: Founded and currently chair the IEEE CIS NNTC Task Force on "Data-Efficient Agentic Learning".
2026.06: Received new funding support from BSC for the project "Training Dynamics of LLMs."
2026.05: Our work on "Agentic DDI" was accepted to KDD 2026.
2026.05: Elected as an ACM Senior Member.
2026.05: Our work on "Dense Communications between LLMs" was accepted to ICML 2026.
2026.04: Our survey on "Data-Efficient Agentic Learning" was accepted to IJCAI 2026.
2026.04: Our work on "Speeding up Mobile GUI Agents" was accepted to IJCAI 2026.
2026.04: Delivered a keynote at the ICLR 2026 Paper Sharing Event hosted by 机器之心.
2026.04: Attended the MOE National Conditions Education Seminar as a university representative.
2026.03: Delivered a keynote at the "AGI-Next" Workshop hosted by AI Time.
2026.03: Delivered a keynote at the 1st Technical Meetup of OpenJiuwen.
2026.03: Our work on "Neural KG Reasoning" was accepted to Artificial Intelligence Journal (AIJ).
2026.03: Received new funding support from Huawei for the project "Efficient Multi-Agent Systems."
2026.03: Will serve as Program Co-Chair of ADMA 2026.
2026.01: Featured in People Weekly (《人民周刊》).
2026.01: Our work on "Pre-training of Low-bit LLMs" was accepted to ICLR 2026.
2026.01: Received new funding support from Huawei for the project "Cross-Device Agents."
2026.01: Received new funding support from BNRist for the project "Domain-Specific Agents."
2026.01: Our work "Searching to Modulate for Cold-Start Recommendation" was accepted to IEEE TPAMI.