BioML Seminar 3.3 - A universal framework for molecular modeling with Brandon Wood

​​​[IN PERSON EVENT IN BERKELEY]

​​​​​Join us for a new seminar from the BioML group in Machine Learning at Berkeley, sponsored by Amplify Partners. This week we're hosting Brandon Wood, a research scientist at Meta FAIR, who'll be talking about advancements in molecular modeling and design with AI.

​Huge shoutout to Berkeley MSEA for helping us organize this week's seminar!

​Title: Exploring the Frontier of Universal Machine Learning Potentials

​Abstract: In the past couple years, there has been tremendous progress towards universal machine learning interatomic potentials for chemistry. Nevertheless, it still remains a huge challenge to have a model that is fast, accurate, and generates physically consistent properties/simulations across very diverse chemistries. In this talk, I will discuss our recent releases of the Open Molecules Dataset (OMol25) and the Universal Model for Atoms (UMA), focusing on the motivation, thought process, and outcomes. Additionally, I will dive into some of the new science that has been enabled, and I will finish with thoughts on the future. 

​Speaker Bio: Brandon is a research scientist on the FAIR chemistry team in San Francisco. His research lies at the intersection of deep learning, chemistry/physics, and large-scale computing. Lately, he has been working on generalizable machine learning potentials and generative models for molecules and materials. Prior to joining FAIR, he was a postdoctoral fellow at NERSC and completed his PhD with Kristin Persson at UC Berkeley.