A Containerized Approach to Data Science

Everybody is talking about how containerization and Kubernetes can enhance the productivity of data science teams and increase the time to insights. But, too frequently, use of containerization requires siloed clusters or data movement due to an inability to access data in the primary cluster from the containers. In this talk, I'll cover how containerization benefits all phases of the data science experience; exploration, training, and deployment. And then we'll dig into some Kubernetized options that allow for stateful data science workflows in containers that work with your data in place. In addition, we'll dig into how to create an end-to-end ML workflow on Kubernetes that is production grade and allows for replicable processes and CI/CD pipelines.