Build an AI platform that survives production. Moving a model from demo to production changes the problem.
Now you need to think about compute, GPU utilization, scheduling, model serving, inference routing, networking, storage, observability, security, cost — and increasingly, how autonomous agents interact with your infrastructure.
KubeCon + CloudNativeCon North America brings those pieces together.
Across four days, explore the open technologies, architectures, and operational practices behind production AI – with the people building and operating them.
Start With a Full Day of AI
Monday, November 9 | CNCF-Hosted Co-Located Events
With an All-Access Pass, your AI experience starts before the main conference begins.
Go deep on the infrastructure required to move AI from development into production.
Explore the technologies and practices behind LLMs, inference, retrieval-augmented generation, agentic systems, and scalable cloud native AI infrastructure.
Spend a day with the developers, contributors, and practitioners advancing the Model Context Protocol and the emerging agentic ecosystem.
Explore real-world implementations, best practices, open standards, and the infrastructure needed to build richer, safer, and more capable AI applications.
An All-Access Pass is needed to attend the events on Monday, November 9
Explore the AI Stack
From GPUs and model serving to agents, gateways, observability, and security, explore the technologies and infrastructure behind production AI.
| AI Area | What You’ll Explore | Session Highlight |
|---|---|
| AI Inference + Model Serving Take AI from model to reliable production service. Explore model serving, inference routing, distributed inference, caching, scaling, and performance. | Composing Inference on Kubernetes: Gateway, Schedule, Serve, Observe — Karena Angell, Red Hat + Sean Horgan, Google |
| GPUs + Accelerated Compute Get more from expensive AI infrastructure. Explore GPU scheduling and sharing, accelerators, capacity planning, workload placement, and utilization. | Fixing GPU Time-Slicing on Kubernetes: Dynamic, Memory-Safe Sharing for AI Inference — Julien Semaan, Kubex |
| Agentic AI + MCP Explore agents, multi-agent systems, MCP servers, agent runtimes, sandboxing, and the infrastructure required to operate autonomous workloads safely at scale. | Running MCP Servers at Scale: Sandboxed Tool Execution on Kubernetes — Harkirat Singh, Freelance + Hrittik Roy, vCluster |
| AI Networking + Gateways Dive into inference-aware routing, AI gateways, model-aware networking, load balancing, and Gateway API patterns connecting models, agents, tools, and users. | Who Ran Up the Inference Bill? AI Traffic Control with KServe and Gateway API Inference Extensions — Lasse Vierow, Freelance |
| AI Observability + Operations Explore OpenTelemetry, Prometheus, tracing, cost attribution, and operational automation to understand what production AI systems are doing — and what they’re costing. | Observability for LLM Inference on Kubernetes — Oscar Castillo, Genius Sports + Jeffry Hernández, Wizeline |
| AI Security + Identity Explore agent identity, workload isolation, authorization, sandboxing, multi-tenancy, and secure execution as AI systems gain access to infrastructure, APIs, tools, and data. | The Agent Has Entered the Cluster: AI Auth 101 for Kubernetes — Jackie Maertens + Nilekh Chaudhari, Microsoft |
AI on the Keynote Stage
The AI infrastructure conversation goes beyond the breakout rooms.
Lin Sun, Solo.io
Keynote: Cloud Native Without Limits: What Can AI Agents Do for Us?
See AI agents, MCP, Kubernetes, and cloud native tooling come together — including how agents could help debug production issues, analyze token spending, and automate operational work.
Abdel Sghiouar, Google Cloud
Keynote: The Identity Crisis: Who is the AI Agent in Your Cluster?
As autonomous agents interact with APIs, databases, tools, and infrastructure, traditional identity assumptions begin to break down. Explore the emerging security and identity challenges of agentic workloads on Kubernetes.
Jessica Forrester, Red Hat
Keynote: Sandbox Your Agents
Explore how cloud native technologies, including Confidential Containers and emerging sandboxing approaches, can help isolate AI agents and their workloads.
Jorge Palma, Microsoft
Keynote: Agentic AI, Assembled in the Open
See how open source and Kubernetes can provide a common platform layer for agentic AI — bringing frameworks, orchestrators, compute, and identity together to deploy, schedule, secure, and observe AI agents using familiar cloud native patterns.
Build What Comes After the AI Demo
Whether you’re figuring out how to serve models efficiently, get more from your GPUs, build an AI platform, operate agents, secure MCP workloads, understand inference networking, or make AI observable, you’ll find the people and open technologies working on the same problems in Salt Lake City.
