264001
eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJpc3MiOiJid2V2ZW50cy1wdWJsaWMtYXBpIiwiYXVkIjpbInB1YmxpYy13ZWIiXSwic3ViIjoiaHR0cHM6Ly93d3cuZ29vZ2xlY2xvdWRldmVudHMuY29tIiwiaWF0IjoxNzc1MDUzNzQxLCJuYmYiOjE3NzUwNTM3NDEsImV4cCI6MTc3NTA1NzM0MSwiY29udGV4dCI6eyJwYXlsb2FkIjp7ImlkIjoiIiwicmVnVHlwZU5hbWUiOiIiLCJldmVudF9pZCI6IjI2NDAwMSIsImNfMzI5MDQxNiI6IiJ9fX0.9wgOgZD54gpjs3JE8eAOAdzR2cMV4RdsuW8cuvq1o4s
phase 5 session info modal
Day 1 – April 22, 2026
Breakouts
BRK3-032 • Kubernetes, Open Models • Advanced Technical
Large-scale LLM inference on GKE
location_on Mandalay Bay H
schedule 1:30 PM - 2:15 PM
Stop overspending on AI infrastructure. High costs and GPU scarcity shouldn’t stall your innovation. In this session, learn how to turn Google Kubernetes Engine (GKE) into a powerhouse for LLM inference. We’ll move beyond basic deployments to master specific optimizations for model loading, accelerator utilization, and smart load balancing. By the end, you’ll know how to architect GKE environments that balance speed and availability while slashing operational costs. Don’t just run models – run them efficiently, reliably, and at scale. Build faster, spend smarter, and maximize every T4, L4, and H100 in your fleet. Read more
This agenda widget contains logic for handling same sessions available at multiple times
GTM Tracking Header name present
Session
contains logic for gtm tracking when modal open