‍

Bodywork v3.0.0 is available to download from PyPI

For version 3.0.0 we've completely reworked the CLI to deliver a more intuitive and responsive user experience, making Bodywork the easiest way to deploy a ML pipeline on Kubernetes. Deploy directly from Git to Kubernetes with a single command:

‍

$ bw create deployment "https://github.com/my-corp/my-pipeline" --branch "main"

We have also added first class support for Jupyter notebooks, that can now be run as stages in a pipeline - just configure them as the executable within a stage and you're all set:

‍

stages:
  train-model:
    executable_module_path: train_model.ipynb

Secrets are now easier to manage using groups - e.g., assign secrets to belong to ‘dev’ and ‘prod’ and then easily switch between the two in `bodywork.yaml`:

‍

pipeline:
  secrets_group: dev

‍

Full details can be found in the official release notes.