About Me

Here is Rui Zhao (Terry, 赵睿).

I am a passionate and self-driven undergraduate student in the Department of Computer Science and Technology at Tsinghua University. My research interests span Reinforcement Learning, World Models and Embodied AI, Computer Vision, and Generative Foundation Models. I have a strong foundation in mathematics and deep learning, with practical experience in LLMs and agents.

Currently, I am working under the guidance of Prof. Alex Lamb, exploring world models and dynamics understanding. I also have experience working with Postdoc Songjie Niu on knowledge graph construction and with Prof. Hongning Wang on AI teaching assistant and reinforcement learning for cognitive reasoning.

I am interested in history, philosophy, literature and music so that I can think about the problems in the world from multiple perspectives. My long-term goal is to integrate high-level symbolic reasoning with low-level embodied perception, working toward generalist agents that can both reason about and physically interact with the open world.

If you are interested in any aspect of me, I am always open to discussions and collaborations. Feel free to reach out to me at - zhaor23 [at] mails.tsinghua.edu.cn

🧭 Thinking About AGI Under Epistemic Humility →
A working framework on what constrains AI, where its real ceiling lies, and how my research fits in. A document I keep revising — if you think any part of it is wrong, I would like to hear from you.

Research Interests

Reinforcement Learning
World Models and Embodied AI
Computer Vision
Generative Foundation Models

My research is driven by the conviction that physical-world dynamics cannot be acquired from human-produced text alone — that the embodied side of intelligence requires environments where learning is not bound by physical time. Following this, I have moved progressively from symbolic knowledge and reasoning toward generative world models and policy learning.

I see the integration of symbolic, linguistic intelligence with embodied, multi-modal perception as one of the central unsolved problems on the path toward general intelligence, and I hope to contribute to it through both architectural innovation in non-language modalities and the exploration of digitized environments as training grounds. Where this thinking currently stands is written down in a working framework.

Skills

Languages: Python, C, C++, Rust, Typescript, Javascript, SystemVerilog
Frameworks: PyTorch, HuggingFace, verl, Weight & Biases, OpenCV, Langchain
Tools: Git, Docker, Linux, LaTeX, VSCode

News and Updates

May 2025: Our work Tree-KG has been accepted at ACL 2025 main conference!
February 2025: Start developing my personal website.

Other Ways to Reach Me

Wechat: xxldzzr
QQ: 2776604248
Academic Email: zhaor23@mails.tsinghua.edu.cn
Main Email: zhaorui0329@gmail.com
Daily Email: 2776604248@qq.com