I am a Senior Research Scientist at NVIDIA Research, where I work on algorithmic data synthesis for reasoning models, spanning inference-time scaling, agents, and reinforcement learning.

For the past decade, one question has driven my research: what data do we need to scale intelligence, and what algorithms can synthesize it? I first asked it in computer vision, through interactive annotation (Polygon-RNN++), learned synthetic data generation (Meta-Sim), and domain adaptation (f-DAL), culminating in my PhD thesis, Visual Learning using Synthetic Data, which developed a framework for visual learning from simulated data. Today I ask it of reasoning models, developing algorithms that synthesize the experience they learn from, on tasks ranging from math, coding, and multimodal understanding (Golden Goose, LongPerceptualThoughts, Long Grounded Thoughts) to scientific reasoning (MagicBeanStalk).

I joined NVIDIA Research as the first full-time employee of the NVIDIA Toronto AI Lab, and later became a founding member of the LLM Reasoning Group (NVR-LACR) led by Yejin Choi. I received my PhD in Machine Learning and Computer Vision from the University of Toronto, advised by Sanja Fidler, and was affiliated with the Vector Institute for AI. I completed my Master’s in Applied Computing at Toronto in 2018 and received the 2020 Microsoft Ada Lovelace Fellowship.

Graduate students interested in internships at NVIDIA are welcome to contact me with a CV and summary of research interests.