264001
eyJ0eXAiOiJKV1QiLCJhbGciOiJIUzI1NiJ9.eyJpc3MiOiJid2V2ZW50cy1wdWJsaWMtYXBpIiwiYXVkIjpbInB1YmxpYy13ZWIiXSwic3ViIjoiaHR0cHM6Ly93d3cuZ29vZ2xlY2xvdWRldmVudHMuY29tIiwiaWF0IjoxNzc3Mzg0MjEyLCJuYmYiOjE3NzczODQyMTIsImV4cCI6MTc3NzM4NzgxMiwiY29udGV4dCI6eyJwYXlsb2FkIjp7ImlkIjoiIiwicmVnVHlwZU5hbWUiOiIiLCJldmVudF9pZCI6IjI2NDAwMSIsImNfMzI5MDQxNiI6IiJ9fX0.KYlPPWkuL0NU5tHeuMqcPJABWUJIAk2tqNJJigFx2pE
phase 7 playlists session info modal
Day 1 – April 22, 2026
Lightning Talks
DEVLT-223 • Gemini, Serverless • Technical
Scaling Gemini Inference
location_on Developer Theater
schedule 3:30 PM - 3:55 PM

Want to process 100k+ Gemini requests now? The Gemini Batch API is designed for this but can take upto 24 hours. What if you need to respond faster? Join this tactical session on building a high-throughput inference engine with Cloud Run. We share real-world architecture on scaling instances, tuning CPU vs. Concurrency, and surviving "429 Rate Limit" storms. Learn to deal with the quotas and limits, optimize max_concurrency, and architect for massive parallelism using serverless containers.

 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