From Chatbots To Agentic AI: Running NVIDIA Dynamo the Kubernetes Way
Talkat KubeCon + CloudNativeCon + OpenInfra Summit + PyTorch Conference China 2026 · 2026-09-07..09-09·
summary
Covers running NVIDIA Dynamo and RoleBasedGroup (RBG) on Kubernetes to support agentic AI inference: using Dynamo for cache-aware routing and disaggregated (prefill–decode) serving, and using RBG for stateful session continuity, discovery, and role/unit coordination across multi-stage multimodal pipelines. Describes deployment patterns (Dynamo-native vs RBG), failure modes under KV-cache contention and partial pod/GPU failures, and Kubernetes-native models for routing, session recovery, and observability.
AI-written summary · about
related talks
19mDisaggregated Intelligence: Scaling AI Agents with Dynamo and Kubernetes
39mPowering Generative AI with Kubernetes: A Cloud Native Approach
27mFrom Chatbots to Agents: Orchestrating Agentic AI on Existing Enterprise Kubernetes Infrastructure.
41mPanel: AI Infra Best Practices: Enterprise Do’s and Don’ts
30mSo You Want To Run AI Agents on Kubernetes: A 101 Guide
32mKeynote: Powering Generative AI with Kubernetes
09mKeynote: Kubernetes for GenAI: The New Operating System of the AI-First World
16mAI + Kubernetes: What Beginners...