Graduate Student · Peking University

HELLO, I'M JATHON

Explore my research GitHub

Peking University · AI

Graduate student in Artificial Intelligence.

OpenClaw · PR #161980

A bug fix merged into a 390k-star open-source AI agent.

About

Curious about how machines learn to think — and how to build the systems that let them.

I’m Jathon, a graduate student in Artificial Intelligence at Peking University, with an undergraduate background in Computer Science from the Harbin Institute of Technology (HIT).

I’m fascinated by the full lifecycle of large models: how pre-training lays the foundation, how post-training and reinforcement learning shape behavior, and how models grow into agents that reason and act in the real world. Alongside research, I enjoy full-stack AI engineering — good ideas deserve good systems.

Now
Graduate Student, AI @ PKU
Previously
B.Eng. in CS @ HIT
Focus
RL · Agents · Reasoning
Based in
Beijing, China

Research

From pre-training to agents — the whole pipeline that turns raw data into models that think and act.

  1. 01

    Pre-training

    Data, scale, and architecture — the choices that set a model's capability ceiling.

  2. 02

    Post-training

    SFT and preference optimization that turn a base model into a genuinely useful assistant.

  3. 03

    Reinforcement Learning

    Policy optimization and reward design that push models beyond imitation.

  4. 04

    Reasoning

    Teaching models to think step by step and solve problems that take many steps.

  5. 05

    LLM Agents

    Models that plan, use tools, and complete long-horizon tasks in real environments.

  6. 06

    Full-stack AI Development

    The engineering underneath it all — taking models from research code to real products.

Open Source

Contributing to the open-source AI ecosystem, one careful fix at a time.

  1. 01

    Fixing approval replay in OpenClaw

    openclaw/openclaw · October 11, 2026

    390k+
    Repository stars
    #161980
    Merged pull request
    Oct 2026
    Merged upstream

Education

Where I've built my foundations.

  1. Peking University

    Graduate Student · Artificial Intelligence

    Present

  2. Harbin Institute of Technology

    B.Eng. · Computer Science

Contact

Into RL, agents, or reasoning? I'm always happy to exchange ideas.

  1. 01

    GitHub

    github.com/Program-123

Questions or ideas? Find me on GitHub.