Towards Consistent AI with Kubeflow
Artificial Intelligence (AI)/Machine Learning (ML) is a strategic direction for enterprises. In the multi-cloud world that enterprises are embracing, there is a need for a robust and consistent AI strategy such that the investments hold good across multiple public clouds, on-premise infrastructure, and all the way to network edge. A CIO needs a consistent AI strategy that would be future proof and avoid vendor lock in. This includes decisions ranging from infrastructure choices all the way to model lifecycle management and safety/compliance. We will first present some challenges that enterprise users face and how Kubeflow addresses some of the pain points. Then we will highlight some of the useful tools within Kubeflow including Kubebench, Katib, and some advanced efforts like AutoML. We will finally conclude with some of the potential directions for the community.