Monitoring models for drift and degradation is not easy - theoretically or practically. We've teamed-up with Aporia to demonstrate how their platform can solve these problems for you, by using their Python client from within a Bodywork pipeline.

Second in a series of articles demonstrating how to engineer and deploy a ML pipeline. This part focuses on developing the pipeline: automating model-training, serving the latest model, handling errors, writing tests and reusing code across projects.

First in a series of articles demonstrating how to engineer a ML pipeline and deploy it to a production environment. This part covers: solution architecture, project structure, setting-up automated tests, making an initial deployment and configuring a CI/CD.

A guide to CICD for ML, together with a template repository to get you started.

bodywork-core v2.1.0 is available to download from PyPI.

Handling prediction uncertainty in production systems using Bayesian inference and probabilistic programs.

Bodywork is flexible enough to deploy almost any type of Python project to Kubernetes. We demonstrate this by using it to deploy a production-ready instance of MLflow, then show how MLflow can be used alongside Bodywork's ML deployment capabilities, to form a powerful open-source MLOps stack.

Get up-and-running with Kubernetes in under 10 minutes from a standing-start.