Author

Josh Romero is a Senior Developer Technology engineer at NVIDIA where he works accelerating HPC and DL applications using GPUs. His work focuses on distributed training performance, especially on large scale HPC computing systems. He is an active contributor and member of the technical steering committee for Horovod, a popular software package for distributed DL training. He was awarded the ACM Gordon Bell Prize in 2018 for applying large-scale deep learning on the Summit supercomputer to analyze extreme weather patterns in climate data. Before joining NVIDIA, he received his PhD from Stanford University where his doctoral research focused on developing computational fluid dynamics solvers for GPUs.