Overview
The Supercompute Observability Software team owns keeping our supercompute centers running reliably at the unprecedented scale required for frontier AI training and inference by building software systems and platforms to automatically predict, monitor, and rapidly respond to changing operational conditions and incidents. Our clusters are large, heterogeneous, and always on - the team makes them observable and keeps them running: we build the infrastructure that lets engineers and technicians expedite and automate their day to day, speeding node remediation and increasing overall GPU uptime.