Polyaxon have a native integration of Kubeflow's components.

We believe that both platforms complement each other, and we are working on making an integration with this community effort as painless as possible.

With Polyaxon's Kubeflow integration, users will have a choice to schedule distributed experiments using our native behaviour or they can leverage Kubeflow’s operators, i.e. TFJob, PytorchJob, and MPIJob in an easy way.