K8s has become utterly essential in cloud computing, but is it getting the same traction in machine learning? Can Kubernetes make machine learning easier, or is it too low-level for data scientists unless it’s abstracted away? What’s the perfect interface for Kubernetes, and how can we make it work for us in AI/ML?
Questions:
• Do data scientists ever need to deal with K8s? Should they or should that always be a separate team?
• Is native k8s enough, or do we need specialized tools layered on top?
• What are some of the top platforms built on top of K8s?
• Has Kubeflow gained any traction or is it falling behind the rest of the industry as a toolset?
• Should companies try to manage their own k8s, or are they better off with a managed service?
• Is Kubernetes necessary for scaling, or are there alternatives that make a lot of sense for industrial-scale AI?