Lisa Lee

Lisa Lee is a Research Scientist at Google DeepMind, focusing on creating AI agents that emulate biological learning and adaptability. She previously taught at Princeton University and received TA awards for Deep Reinforcement Learning and Probabilistic Graphical Models.

• Education: PhD in Machine Learning from Carnegie Mellon University (advised by Ruslan Salakhutdinov and Eric Xing); A.B. in Mathematics from Princeton University (advised by Sanjeev Arora).

Her research centers on AI embodiment, intrinsic motivation, and hierarchical planning. She explores how evolutionary-inspired inductive biases and memory mechanisms can enable agents to generalize across physical and conceptual domains, as demonstrated in her work on robotic agility benchmarks and multimodal transformers.

Notable scientific contributions include the Barkour quadruped robot benchmark, Gemini multimodal models, and theoretical work on causal language models. She co-organized key AI workshops at NeurIPS and ICML, and her awards include Princeton's TA of the Year for technical courses.

• Leadership: ICML Workflow Chair (2019), NeurIPS workshop co-organizer (2019, 2021), peer reviewer for top AI conferences.