Photo of Yang Sheng

Yang Sheng 盛杨

Ph.D. Student

Fudan University & Shanghai Innovation Institute

I am a first-year Ph.D. student broadly interested in understanding and advancing artificial intelligence toward safe and capable AGI. My vision is to see AI systems that are both powerful and trustworthy — systems that can genuinely propel human civilization forward. To this end, I focus on mechanistic interpretability to uncover the internal workings of large language models, and explore approaches in continual learning and reinforcement learning to enable autonomous self-improvement. I believe that deeply understanding how AI thinks is the foundation for building AI that humanity can rely on. I am also interested in building practical AI tools, technologies, and products with real-world impact, and I am always happy to connect with people who share similar interests or would like to collaborate.

Large Language Models Mechanistic Interpretability Continual Learning Reinforcement Learning

News

Publications

2026

FormalJudge: A Neuro-Symbolic Paradigm for Agentic Oversight

Jiayi Zhou*, Yang Sheng*, Hantao Lou*, Yaodong Yang, Jie Fu (* equal contribution)

arXiv preprint arXiv:2602.11136, 2026

Selected Projects

Portfolio Daily Tracker

Self-hosted investment portfolio tracking and AI trading assistant

2026

A full-featured tool for tracking and analyzing multi-market portfolios across A-shares, Hong Kong, and US stocks. It combines daily reporting, backtesting, portfolio analytics, and AI chat support in a self-hosted workflow-oriented system.

Python React AI Tooling Quant Workflow

Education & Experience

Contact

Office

Shanghai, China