|
Zhaoyu Li (李照宇)
I am a Computer Science Ph.D. student at the University of Toronto (U of T), working with
Xujie Si and Kaiyu Yang (at Meta).
Before joining U of T, I spent one and a half years as a Ph.D. student at McGill University
and Mila - Quebec AI Institute.
Prior to that, I obtained my bachelor's degree with honors from the ACM class at Shanghai Jiao Tong University (SJTU).
Email /
CV /
Google Scholar /
X /
GitHub
|
|
Research Interests
I am broadly interested in deep learning and automated reasoning.
My recent work focuses on AI for mathematics, where I train large language models
to generate mathematical proofs in both formal and natural-language settings.
|
Work Experience
-
Research Scientist Intern, Meta FAIR
May 2025 - September 2025, New York, NY, USA
-
Part-Time Student Researcher, Meta FAIR
September 2025 - December 2025, Toronto, ON, Canada
|
Selected Papers
* denotes equal contributions. See full list on Google Scholar.
-
Euclid-Omni: A Unified Neuro-Symbolic Framework for Geometry Problem Solving
In submission
Zhaoyu Li, Hangrui Bi, Youyuan Zhang, Wenjie Ma, Zenan Li, Xujie Si, Kaiyu Yang
-
PyEuclid: A Versatile Formal Plane Geometry System in Python
[paper]
[code]
CAV 2025
Zhaoyu Li*, Hangrui Bi*, Jialiang Sun*, Zenan Li, Kaiyu Yang, Xujie Si
-
Proving Olympiad Inequalities by Synergizing LLMs and Symbolic Reasoning
[paper]
[code]
ICLR 2025
Zenan Li*, Zhaoyu Li*, Wen Tang, Xian Zhang, Yuan Yao, Xujie Si, Fan Yang, Kaiyu Yang*, Xiaoxing Ma*
-
Autoformalizing Mathematical Statements by Symbolic Equivalence and Semantic Consistency
[paper]
[code]
NeurIPS 2024
Zenan Li, Yifan Wu, Zhaoyu Li, Xinming Wei, Xian Zhang, Fan Yang, Xiaoxing Ma
-
A Survey on Deep Learning for Theorem Proving
[paper]
[code]
COLM 2024
Zhaoyu Li, Jialiang Sun, Logan Murphy, Qidong Su, Zenan Li, Xian Zhang, Kaiyu Yang, Xujie Si
-
Autoformalizing Euclidean Geometry
[paper]
[code]
ICML 2024
Logan Murphy*, Kaiyu Yang*, Jialiang Sun, Zhaoyu Li, Anima Anandkumar, Xujie Si
-
G4SATBench: Benchmarking and Advancing SAT Solving with Graph Neural Networks
[paper]
[code]
TMLR 2024
Zhaoyu Li, Jinpei Guo, Xujie Si
-
Learning Reliable Logical Rules with SATNet
[paper]
[code]
NeurIPS 2023
Zhaoyu Li, Jinpei Guo, Yuhe Jiang, Xujie Si
-
NSNet: A General Neural Probabilistic Framework for Satisfiability Problems
[paper]
[code]
NeurIPS 2022
Zhaoyu Li, Xujie Si
-
Graph Contrastive Pre-training for Effective Theorem Reasoning
[paper]
ICML Workshop on Self-Supervised Learning for Reasoning and Perception 2021
(Contributed talk)
Zhaoyu Li, Binghong Chen, Xujie Si
|
Selected Talks
-
PyEuclid: A Versatile Formal Plane Geometry System in Python
Modeling Reading Group, U of T, 2025
-
Deep Learning for Theorem Proving: Overview and Future Directions
Applied Deep Learning Research Group, NVIDIA, 2024
-
Deep Learning for Theorem Proving: Overview and Future Directions
Software Engineering Group, Nanjing University, 2024
-
Some Notes on Neural Theorem Proving
Systems and Networking Research Group, Microsoft Research Asia, 2023
|
Selected Awards
- DIDI Graduate Student Awards, U of T, 2024
- Max Stern Recruitment Fellowship, McGill University, 2021
- China National Scholarship (Top 0.2%), SJTU, 2020
|
Teaching Experience
- Teaching Assistant, CSC2547: Automated Reasoning with Machine Learning, 2023, 2024, U of T
- Guest Lecturer, MS326: Deep Learning and Its Applications, 2022, SJTU
- Teaching Assistant, MS110: Computer System (2), 2020, SJTU
- Teaching Assistant, MS108: Computer System (1), 2019, SJTU
|
Academic Service
- Conference Reviewer: NeurIPS, ICLR, ICML, COLM
- Journal Reviewer: TMLR, Nature Machine Intelligence
|
Miscellaneous
In my spare time, I enjoy reading history and detective fiction, working out at the gym, and playing badminton and table tennis.
In earlier years, I was also quite good at soccer and MOBA games — I reached 8000+ MMR (TOP 1000) in Dota 2.
I also enjoy traveling and love picking up new sports whenever I get the chance.
|
|
Website template adapted from
Jon Barron
Last updated: November 2025
|
|