I'm Zhicheng Jiang
MIT undergraduate researching generative models, automated theorem proving, and mathematical reasoning.
you@my_world: ~$ df -h
| Stage |Total|Used|Free|Used%| Mounted On
Baby 10Y 10Y 0 100% /CHN/Guangdong/Shenzhen
Middle School 5Y 5Y 0 100% /CHN/Guangdong/Shenzhen/SMS
Pre-College 1Y 1Y 0 100% /CHN/Beijing/THU/IIIS
Undergrad 4Y 2Y 2Y 50% /USA/MA/MIT
Future inf 0 inf 0% /
you@my_world: ~$ jzc get-email
jzc_2007@mit.edu
you@my_world: ~$ jzc get-phone -n
8576390768
you@my_world: ~$ jzc get-future-plan -y 3
Learn and explore more at MIT!
you@my_world: ~$ jzc get-future-plan -work
Build systems that help AI reason.
you@my_world: ~$ exit
bye-bye!
logout
Connection to 114.514.1919.810 closed.
About Me
I am Zhicheng Jiang, an MIT Class of 2028 undergraduate double-majoring in Mathematics and Computer Science.
After a preparatory year at Tsinghua University, I developed a strong interest in artificial intelligence and machine learning. I am especially interested in turning human intuition into mathematical formulations, algorithms, and models that solve real problems.
I competed in the IMO in high school and remain interested in AI for mathematics—and, more broadly, how to build systems that reason.
At MIT, I work with Kaiming He on generative models. I have also worked on automated theorem proving with ByteDance Seed AI4Math and on agentic data pipelines at Abaka AI. I welcome conversations and collaborations with people who share these interests.
Besides academics, I also like playing table tennis and badminton, and all kinds of strategy games. I am always welcome to make new friends and have fun together!
For a concise overview, view my resume (PDF).
AI Research
Exploring generative models and mathematical foundations of AI.
Mathematics
IMO Gold Medalist with strong interest in mathematical thinking.
Programming
Experienced in C++ and Python with focus on ML frameworks.
Education
Massachusetts Institute of Technology
Undergraduate, Double Major in Mathematics and Computer Science
Core Courses:
- 6.S978(G) Deep Generative Models (2024 Fall)
- 6.8611 Quantitative Methods for Natural Language Processing (2024 Fall)
- 6.4110 Representation, Inference and Reasoning in AI (2025 Spring)
- 6.6410 Quantum Computation
GPA: 5.0/5.0
Tsinghua University (IIIS, Yao Class)
Preparatory Year
Core Courses:
- Linear Algebra (2023 Fall)
- Advanced Calculus (2023 Fall)
- Algorithm Design (2023 Fall)
- Intro to Scientific Research of Lab (2023 Fall)
- Intro to Programming in C/C++ (2023 Fall)
- Abstract Algebra (2024 Spring)
- Intro to Computer Systems (2024 Spring)
- Deep Learning (2024 Spring)
- Intro to LLM Applications (2024 Spring)
GPA: 4.00/4.00
Shenzhen Middle School
High School Education
Experience
Research Engineer, Abaka AI
Built agentic systems to improve data pipelines and conducted independent research.
ByteDance Seed AI4Math
Developed algorithms for Seed-Prover, an agentic automated theorem prover that earned a silver-medal-level result at IMO 2025 and achieved state-of-the-art results on PutnamBench and other challenging mathematics benchmarks.
Undergraduate Researcher, MIT
Researching diffusion and flow-matching models with Kaiming He, with an emphasis on the principles underlying denoising-based generative models.
Undergraduate Researcher, Tsinghua University
Worked with William Kuszmaul on theoretical analysis and algorithm design for randomized data structures.
Skills
Programming Languages
Frameworks & Libraries
Languages
Awards
Top 25
William Lowell Putnam Mathematical Competition (2025)
Gold Medal
64th International Mathematical Olympiad (IMO 2023)
First Prize
National Olympiad in Informatics in Provinces (NOIP 2021)
Projects
Deep Learning Course Project focused on transferring artistic styles to video content while maintaining temporal consistency.
LLM Applications Course Project exploring the integration of large language models with database systems.
Deep Generative Models course project. A new method to speed up diffusion models by one-step generators. Read more in the blog post.
Publications
Is Noise Conditioning Necessary for Denoising Generative Models? (ICML 2025)
This research explores the necessity of noise conditioning in denoising generative models, investigating alternative approaches and their implications for model performance and efficiency.
Seed-Prover: Deep and Broad Reasoning for Automated Theorem Proving
An automated formal theorem prover that achieves state-of-the-art performance on olympiad-level benchmarks including PutnamBench.