Room: Salon H Since last year's swampUP, the Cloud Native AI ecosystem has evolved significantly. With ongoing innovation in open source models, orchestration, model registries, and observability tools, DevOps teams are now better equipped to reliably and quickly move AI applications from experiment to production. This talk will dive into the latest work from the CNCF Cloud Native AI Working Group, which has been mapping out the reference architectures, challenges, and opportunities at the intersection of MLOps, DevOps, Security and cloud native tooling. We'll showcase the most recent advancements across the model lifecycle—covering artifact management, inference serving, workload orchestration, and lineage tracking—through a lens of practical, real-world implementation. Attendees will leave with a blueprint for building scalable, reproducible, secure AI platforms using cloud-native practices, including open-source components. We'll also explore where the ecosystem is heading next, from AI gateway patterns to cluster-level scheduling for GPUs.