Dai (Ethan) Shi
Postdoctoral Research Fellow · University of Cambridge
I work with José Miguel Hernández-Lobato at the University of Cambridge on theory-grounded machine learning: generative models, reinforcement learning, stochastic analysis, operator learning, graph neural networks, multimodal LLMs, multimodal pretraining and mechanistic interpretability.
Before Cambridge I was at the University of Sydney as a research fellow with Professor Junbin Gao, 2023–2025. I was also a member of China's reserve team for the International Mathematical Olympiad (IMO).
Open to industry research roles. I am looking for research scientist roles in machine learning and scientific ML, and quantitative researcher roles. Email is the quickest way to reach me.
Cambridge, United Kingdom
Watch · Recent paper
Do AI models see the same world?
A short video on our new paper, From Plato to Aristotle: Toward a General Theory of Representation Hypotheses. Studies disagree on whether different AI models converge to the same representations; we show that their alignment criteria sit at different points of one lattice.
Research
Each theme links to its key papers below.News
Service & Recognition
Recognition
- Winning team (team lead), Singular Stochastic PDE Learning Competition, 2025 Learning solutions of the dynamical Φ42 model · competition page
- Member of China's reserve team, International Mathematical Olympiad
Service
- Organizer & Program Chair NeurIPS 2026 Workshop on AI for Stochastic Dynamics: From Theoretical Foundations to Scientific Applications · Sydney, 11 December 2026
- Conference reviewer / Area Chair ICLR, ICML, NeurIPS, KDD, VLDB
- Journal reviewer Artificial Intelligence, JMLR, TMLR, Machine Learning