I am a Research Scientist at Meta FAIR, where I develop generative AI methods that accelerate scientific discovery. My work spans generative modeling, model scaling, and building large-scale open scientific benchmarks, with the goal of making computational science faster and more accessible.

My current research focuses on AI for Science. I lead efforts on universal machine-learning interatomic potentials (UMA), generative models for crystal structure design (FlowLLM, FlowMM), and some of the largest open datasets at the intersection of AI and materials science. Together, these efforts accelerate materials discovery by replacing expensive quantum-mechanical simulations with fast, accurate AI models.

I led fastMRI, a collaboration between Meta AI and NYU Langone to accelerate MRI scanning using AI. Our methods achieve up to 4x acceleration with no loss in diagnostic accuracy and have become the clinical standard for accelerated MRI worldwide.

Previously, I built and led the speech research team at FAIR, where we trained some of the earliest multi-billion parameter speech models and developed self-supervised methods that served billions of users across Meta's products. Before joining Meta, I was part of the team at Baidu that built Deep Speech 2, one of the first end-to-end neural speech recognition systems.

I hold a Master's in Language Technologies from Carnegie Mellon University, and my research has been widely featured in the popular press.