Project Description

So you have an idea for a machine learning project for HackWeek. Have you thought about what tools you'll be using? Choosing the right set of machine learning tools and making them work together can be time consuming, not to mention the unavoidable learning curve. Perhaps you could use some help with that.

The SUSE AI/ML team has the answer: FuseML - an open source machine learning DevOps orchestrator that can get your machine learning projects up and running as easy as lighting a fuse.

FuseML started as a spin off project Carrier. Think "Carrier for Machine Learning": you write your ML application using one of the popular machine learning libraries (e.g. scikit-learn, TensorFlow, PyTorch, XGBoost) and FuseML takes care of all operations necessary to get your machine learning models in action, so you can concentrate on your code.

The catch: FuseML is still in a pre-alpha state, although it can already be used to showcase basic features. While using it, you may run into some corner cases we haven't covered yet, but you'll not be alone: we're here to help.

The rewards: access to expert knowledge in AI/ML and a chance to have your ML project published into the FuseML gallery of sample applications.

What you'll need: to install and use FuseML, you'll need a kubernetes cluster. If you don't already have one handy, or if you're low on hardware resources, you can install minikube, kind or k3s on your machine.

Goal for this Hackweek

• discover new use cases and AI/ML tools to be enabled for FuseML
• offer assistance and guidelines on AI/ML best practices and tools in the context of FuseML
• pimp up FuseML's gallery of sample applications

Resources

• FuseML github project page
• RocketChat channel: #machine-learning