Farhad Ramezanghorbani

Farhad Ramezanghorbani is a senior research scientist at NVIDIA, where he focuses on the deployment and acceleration of large language models (LLMs) for multiomics, including protein and genomic applications. His research explores alternative attention mechanisms with subquadratic scaling to enable ultra-long context modeling of biological sequences, advancing the capabilities of AI in life sciences. Prior to joining NVIDIA, he was a Senior Scientist in Machine Learning at Schrödinger, where he developed neural network potentials and methods for drug discovery. He has also contributed to widely used open source projects such as NeMo, BioNeMo, TorchANI, PyTorch, and QCEngine. Farhad holds a PhD in Chemistry from the University of Florida, where his research centered on neural network potentials, active learning, and coarse-grained biomolecular modeling. His broader interests span AI for scalable deep learning systems for molecular science in drug and material discovery.