From Bash Scripts to Kubeflow and GitOps - Operationalizing ML at Scale

Explore the journey of operationalizing data science workloads at scale using Kubeflow and GitOps in this 34-minute conference talk from CNCF. Learn how DHL Data & Analytics transitioned from bash scripts to a standardized project delivery approach with Kubeflow, improving developer onboarding and solutions support. Discover the MLOps architecture featuring Kserve for API creation and Kubeflow pipelines for batch predictions delivery. Gain insights into using Kubeflow as a platform for reproducible ML experiments and understand the importance of diverse professional experiences in supporting scale-up for growing teams. Examine the GitOps approach and its integration within the Kubeflow ecosystem, including tracking objects in Git repositories and implementing complete project deployments using GitHub actions and ArgoCD.

From Bash Scripts to Kubeflow and GitOps: Our Journey to Operationalizing ML at Scale