Kubeflow Foundation is a three-day course that helps you take machine learning workflows from ad hoc to production grade. The course gives participants a comprehensive introduction to Kubeflow, a framework for deploying machine learning workflows on Kubernetes. Day one grounds attendees in Kubeflow and Kubernetes fundamentals, and they deploy and configure the platform themselves. The day then moves into data engineering, integrating storage with Kubernetes and using the Spark Operator to run distributed data processing jobs.
Day two turns to model development, opening with Feast, where attendees manage features and metadata through a dedicated feature store. From there, attendees work directly in Kubeflow Notebooks. They build and manage Kubeflow Pipelines to automate multi-step ML workflows. The day closes with the Model Registry, where trained models are versioned and tracked.
Day three covers inference and monitoring, starting with hyperparameter tuning using Katib to search for optimal model configurations automatically. Attendees then train models with the Training Operator before serving them through KServe’s serverless model deployment. The course finishes with monitoring models in production for performance and drift. Lecture and discussion get equal time with hands-on lab work, giving attendees direct experience with every core component. This spans data preparation through model development, training, serving, and monitoring. Upon completion, students are ready to begin designing and working with production-grade MLOps solutions.
Who Should Attend
Developers, Data Scientists/Engineers, ML Engineers, Dev/AI/MLOps professionals
What Attendees Will Learn
Upon completing Kubeflow Foundation, participants will be able to:
- Gain an in-depth understanding of Kubeflow, its architecture, deployment and use
- Explore Cloud Native data engineering techniques in support of MLOps
- Acquire hands-on experience configuring and managing Kubeflow and its components
- Learn how to use Kubeflow Notebooks and Pipelines to build machine learning workflows
- Work with the Kubeflow Feature and Model Stores
- Understand how to use Kubeflow to train and deploy models with the Training Operator and KServe
- Develop the skills and expertise needed to transform ML workflows into production grade end-to-end solutions
Prerequisites
Attendees should basic familiarity with machine learning, familiarity with containers, Kubernetes, and basic Linux command-line skills.