Keynote: Scaling Model Training with Volcano: iFlytek’s Kubernetes Breakthrough
Talkat KubeCon + CloudNativeCon China 2025 · on Tuesday, 2025-06-10· at 09:55· Level 16 | Grand Ballroom I·
summary
Describes how iFlytek used the Volcano scheduler on Kubernetes to address low GPU utilization, fragile workflows, and cross-team resource contention when training large AI models. Explains their approach of intelligent scheduling and orchestration that raised GPU usage by over 40% and reduced failure recovery time by 70%, improving performance, reliability, and fairness on a multi-tenant AI platform.
AI-written summary · about
related talks
37mImproving Performance of Deep Learning Workloads With Volcano
30mVolcano: A Unified Scheduling Platform for Cloud Native AI
43mRun Your AI Workloads and Microservices on Kubernetes More Easily and Efficiently...
30mCloud Native AI: Harness the Power of Advanced Scheduling for High-Performance AI/ML Training
08mNext-Gen AI Orchestration With Volcano On Kubernetes
32mVolcano: Intro & Deep Dive
26mIntro: Kubernetes SIG
06mSponsored Keynote: Building Your AI Factory: Wh...
