David Brandfonbrener

About

David Brandfonbrener’s research focuses on understanding the foundations of modern machine learning. His PhD thesis focused on building principled approaches to the offline reinforcement learning problem. Now he is primarily working at the intersection of foundation models and decision making, spanning problems throughout the learning pipeline, from data collection through finetuning and generation.

Prior to coming to Harvard University, Brandfonbrener completed his PhD at NYU, advised by Joan Bruna and supported by an NDSEG fellowship. During that time, he interned at Google Brain, MSR, and FAIR. Preceding that, he completed his undergraduate degree at Yale University.

Research Focus

Brandfonbrener’s current research focuses on reinforcement learning, language models, and science of deep learning. He uses scalable approaches to solving problems in control, and focuses on decision-making based on deep learning.