DevConf.US 2021

Managing software is a challenging endeavor. Manual operations, fragile snowflake pipelines, and managing deployments across hybrid cloud topologies are among the typical headaches that service-owning teams are all too familiar with. While Kubernetes excels at running microservice applications, the deployment and ongoing management of more complex applications can be time consuming for end users. With an abundance of possible avenues to take, teams are often left questioning what to do. Over the course of this session we will show how we transitioned from the wild west of manually created Kubernetes objects to an easily maintained self-service AI/ML platform managed by an automated continuous deployment solution. We will then demonstrate how to go from an empty cluster to a self-sufficient ML platform with only a few clicks! Attendees will walk away understanding how GitOps can improve reliability and time-to-value for all their hybrid cloud applications, including ML workflows.