docker-stacks

Jupyter Docker Stacks is a collection of pre-built container images that come with Jupyter already installed and ready to use. Jupyter is an interactive computing environment used by data scientists, researchers, and analysts to write code, run it, and see the results in the same document. The Docker images package everything needed to get a Jupyter environment running with a single command, without installing Python packages or configuring a server manually.

The images are published on Quay.io and are tagged by date, so you can pin your setup to a specific snapshot or use the latest version. Several image variants are available, ranging from a minimal base image to a scipy-focused image (with mathematics and science libraries), a data science image (which adds R and Julia alongside Python), and others. A CUDA-enabled variant for GPU computing is also available for x86_64 machines. All images are built for both the common x86_64 architecture and the ARM-based aarch64 architecture, so they work on Apple Silicon machines and cloud servers with ARM processors.

To start a session, you run a one-line Docker command that launches a container and opens a port on your machine. You then visit a URL in your browser to access the JupyterLab interface. You can mount a folder from your computer into the container so your files are accessible inside the session. When the container stops, files in that mounted folder are preserved; everything else in the container is discarded. JupyterLab is the default interface, though you can switch back to the older Jupyter Notebook interface via an environment variable.

The project is maintained by the Jupyter organization. Questions and community discussion happen through the Jupyter Discourse forum. Full documentation lives on ReadTheDocs and covers selecting the right image variant, customizing images for specific workflows, and building your own Dockerfile on top of a stack image.