Building Resilient AI Services Using Multi-Cluster Kubernetes

Discover how to design and operate resilient multi-cluster Kubernetes platforms specifically for AI workloads in this 42-minute webinar presented by Mirantis and Tigera experts. Explore a practical reference architecture where training, inference, and shared services operate in separate clusters to isolate GPU resources, reduce blast radius, and enable independent scaling of each layer. Learn how Mirantis k0rdent AI provides platform teams with a unified control plane for provisioning clusters, applying guardrails, and managing lifecycle operations across environments. Understand how to standardize AI infrastructure using reusable templates for cluster builds and Day-2 operations while enforcing governance for multi-tenancy, quotas, and compliance. See demonstrations of Calico Cluster Mesh and federated services that enable secure, private east-west connectivity and consistent service discovery across k0rdent-managed clusters. Examine zero-trust microsegmentation with Calico policies for controlling cross-cluster traffic and utilizing Calico observability tools for validating flows and troubleshooting issues across clusters and namespaces. Gain practical insights from Neil Overmon, Sr. Director of Solution Architecture at Mirantis, and Aadhil A. Majeed, Sr. Product Manager at Tigera, as they present live demonstrations using the Lens K8s IDE and provide a comprehensive blueprint for building AI-ready multi-cluster platforms that are more secure, reliable, and easier to operate.