Dan Lee (종민)

I’m Dan (종민) Lee from Long Island, currently based in NYC. I’m interested broadly in machine learning and artificial intelligence.

Recently in my research, I’ve thought about the question of general intelligence. Specifically, how can models learn representations that are transferable to diverse domains and learn continually? My current work explores synthetic data as a means to learn a universal representation or computational priors to adapt across domains.

Outside of research, I love cooking, hiking, working out, playing TFT, and reading. I’m particularly drawn to literature that explores identity, abstraction, and meaning.

I’m currently a first-year research-based funded MS student at the TRAIS Lab at UIUC, advised by Professor Jiaqi Ma. I’m also a research collaborator at MIT CSAIL with Professor Pulkit Agrawal. I received my B.S. in Computer Science (Intelligent Systems) with a minor in Economics from Columbia University, where I graduated summa cum laude as an Egleston Scholar. Previously, I was an early-stage technology and software investor.

Life Updates & Achievements

• Training Language Models via Neural Cellular Automata accepted to NeurIPS 2026
• started as a first-year research-based funded MS student at the TRAIS Lab at UIUC, advised by Professor Jiaqi Ma
• graduated summa cum laude as an Egleston Scholar from Columbia University

Pre-Prints & Publications

• Pre-Pre-Training Before Pixels and Genes by Dan Lee*, Seungwook Han*, Nathan Breslow, Mahler Revsine, Daniel Khashabi, Pulkit Agrawal. Spotlight Presentation, NeurIPS 2026 Workshop on Linguistic Principles for Foundation Models.
• Convergent Emergence of In-Context Learning Across Modalities by Nathan Breslow, Seungwook Han, Daniel Hyunsoo Lee, Aayush Mishra, Anqi Liu, Daniel Khashabi. Preprint.
• Training Language Models via Neural Cellular Automata by Dan Lee*, Seungwook Han*, Akarsh Kumar, Pulkit Agrawal. NeurIPS 2026.

Please feel free to email me.