Disclaimer: I work for Microsoft, but I'm not here to tell you to use Azure - just providing an example for the use of AAD. Kubeflow is an OSS framework for do

Disclaimer: I work for Microsoft, but I'm not here to tell you to use Azure - just providing an example for the use of AAD. Kubeflow is an OSS framework for doing distributed data science / machine learning (including Tensorflow hence the name) on Kubernetes. With the AAD integration it's just much more convenient to give a team of data scientists access to a Kubeflow Kubernetes cluster - AAD with the Kubeflow OpenIDConnect component will handle AuthN & AuthZ. Kubeflow itself sets up different Kubernetes namespaces for each authorized user upon first login. You definitely don't need to run Kubernetes on AKS or Azure in general to take advantage of the AAD features.