🦫 Beaver • MLOps for (online) machine learning

Beaver is...

✔ The whole package • it's a framework to develop, deploy, and maintain machine learning models. And that includes feature engineering.

✔ No fuss • there's an SDK to do stuff, and a UI to see stuff.

✔ Online-first • it is designed for online machine learning models, while also supporting batch models.

✔ Opinionated • it encourages you to process data with SQL and define models in Python.

✔ Interfaces all the way down • you can plug in your existing message broker, stream processor, model store, etc. This is called a hexagonal architecture in software design.

✔ Batteries included • default infrastructure and monitoring are provided if needed.

🤱 Getting started

There is a pre-built image you can pull and run 🐳

docker pull ghcr.io/online-ml/beaver:latest
docker run ghcr.io/online-ml/beaver

You can also build the provided docker-compose.yaml yourself:

git clone git@github.com:online-ml/beaver.git
cd beaver
docker-compose up

Go to http://localhost:8501 to access the user interface.

This is all you need to run Beaver. Check out the examples to see how to use it.

👀 Examples

• 🚕 Taxis
• 🌳 Batch gradient boosting vs. random forest

📝 License

Beaver is free and open-source software licensed under the 3-clause BSD license.