FLOps is an Oakestra Addon that enables practical application-oriented federated machine learning.

FLOps automates ML/FL duties for the user by building on top of DevOps and MLOps concepts, including:

• Handling the setup and configuration of the ML environment by building multi-platform container images.
• Scheduling and handling FL learners and aggregators - performing the FL training.
• State-of-the-art observability capabilities via a modern browser-based GUI.
• Providing the trained model for pulling and automatic deployment as an inference server.
• Allow users to customize their FL by specifying various aspects ranging from training rounds to their FL strategy.

📚 Check out the FLOps Wiki to get started.

Requirements

• Linux (preferably Debian/Ubuntu)
• Python Version 3.10+