Wayne Hendricks
I build supercomputers and the training infrastructure behind foundation models.
What I’m Up To Lately#
GPU systems on GCP for clinical-grade computational pathology models at Tempus. Currently migrating Azure training infrastructure to GCP.
Past systems#
Paige.AI, 2022 – 2025. “Azul”, a scalable Azure CycleCloud Slurm cluster with hundreds of GPUs and 7 PB of multi-modal data on tiered storage. Production training infrastructure behind Virchow, Virchow2, PRISM, PRISM2, and OmniScreen.
Memorial Sloan Kettering Cancer Center, 2018 – 2022. “Genesys”, an on-prem AI/ML cluster with 100 GPUs and 5 PB of pathology images in tiered storage. The first cluster with the newly released NVIDIA DGX-1 Volta GPU server platform at a cancer center, a collaboration between the Fuchs Lab and their newly founded spinoff, Paige.
Caltech High Energy Physics, 2015 – 2018. “CMST2”, a 7,300-slot HTCondor cluster with 4.5 PB of physics collision data in HDFS and Ceph. One of the seven major U.S. Tier-2 sites on the Worldwide LHC Computing Grid and the Open Science Grid, the largest global supercomputing infrastructure ever built.