About

I'm a senior research scientist working on agentic AI for science and engineering at NVIDIA's Learning and Perception Research (LPR) lab in Toronto, with Jan Kautz. Before that, I worked on generative physical AI at NVIDIA's Spatial Intelligence Lab (SIL) with Sanja Fidler, on everything from text-to-3D generation to real-time world models for autonomous driving. I completed my PhD in machine learning in 2024, advised by David Duvenaud. The bottleneck in modern AI is no longer raw capability but direction: what a system optimizes for, and how anyone can tell whether it got there. I work on the rewards, verifiers, and evaluation that let models improve themselves, and on the large-scale nested optimization that makes them trainable. Recent work:

Previously, I worked at Google on the production AutoML pipeline and at Facebook (now Meta) AI Research with Jakob Foerster on multi-agent learning. My M.Sc.A.C. at the University of Toronto was a joint program in Statistics and Computer Science focused on Data Science, after which I joined a boutique hedge fund. Even earlier, my undergraduate studies included a double major in Computer Science and Mathematics and a minor in Economics, while working with Dmitry Krass on Operations Research.

Potential Interns: I mentor interns in the LPR lab. See open positions, and email me if your interests overlap with the work above.

E-mail: lorraine@cs.toronto.edu

Location: Toronto, Canada

Papers

Teaching