Tectonic shifts across the industry have created the need for companies to rise to the challenge of running AI at scale, especially the ever-increasing demands on Systems and Reliability. Surupa Biswas and Peter Hoose, leaders of Meta’s Core Infrastructure and Production Engineering teams, will share a keynote talk with learnings and innovations around some of the industry’s hardest challenges that are impacting all levels of the stack. The keynote will highlight exciting stories about some particularly tough topics including capacity for DCs, distributed inference for Llama, supercomputing through a single controller and much more.