Overview

What does Kubeflow Workspace offer?

Kubeflow Workspaces provides on-demand, containerized development environments that run directly on your Kubernetes cluster. This bridges the gap to give data scientists, machine learning engineers and agents a convenient access to the infrastructure in your Kubernetes cluster to code, explore data, and run experiments, while platform administrators keep full control over the images, hardware resources, and security that back those environments.

Kubeflow Workspaces is the next generation of Kubeflow Notebooks, replacing the original Notebooks controller with a more flexible and extensible architecture.

Core Features

• Interactive development environments on Kubernetes: run your IDE next to your data and powerful compute instead of locally on your workstation.
• Support for any web-based IDE, with first-class examples for popular examples such as JupyterLab, Visual Studio Code (code-server), and RStudio.
• Persona-based design Cluster admins curate the available environments through the WorkspaceKind custom resource, including base images, default resources, and pod-level configuration, so endusers can create Workspace resources in a guided wizard without having to understand the underlying Kubernetes resources.

Get Started

Ready to try Kubeflow Workspaces? Follow the Deployment Guide to install it alongside the Kubeflow Community Distribution on a staging or development cluster.

Next Steps

• Read the Deployment Guide to install Kubeflow Workspaces.
• Want to help build it? See Contribute to Kubeflow Workspaces.
• Explore the current Kubeflow Notebooks documentation.
• Follow development in the kubeflow/notebooks repository on GitHub.