AWS re:Invent 2019: Building machine-learning infrastructure on Amazon EKS with Kubeflow (CON306-R1)

Published on Dec 05, 2019

Data science, machine learning (ML), and artificial intelligence have exploded in popularity in the last few years, with companies building out dedicated ML teams. Kubeflow is the ML toolkit for Kubernetes and provides useful components resolving problems in multiple areas. For enterprise users, it still cannot fulfill all the requirements for out-of-the-box deployment. Some customizations need to be made, like authentication, training job monitoring, user profile, and resource quota management. In this session, we show you a real use case of how to build ML using Kubeflow, along with challenges and solutions to address them.