The FEDn framework

Streamlining federated learning with seamless integration, flexible deployment, and instant start.
Release notes »
Overview
FEDn is an open-source federated machine learning framework. The FEDn SDK, free from IP-risk or lock-in, features secure communication protocols, efficient deployment tools, and plugins for functionalities like distributed logging and model management. It facilitates straightforward backend deployment and on-premise installation.

Addressing enterprise requirements, we developed FEDn Studio, enhancing FEDn with advanced features for accelerated development and seamless production transition. FEDn supports deployment on any infrastructure, including Kubernetes.
The FEDn framework overview

Framework features

FEDn is designed for scalability with many clients, is enterprise-ready with authentication and security features, supports various servers and devices, integrates easily, works with industry-standard technology, enhances usability for MLOps, and is compatible with multiple machine learning frameworks.
Scalability & Resilience
FEDn was built to enable real-world systems with a large number of clients and to recover from failure in all system components.
Enterprise ready
Authentication and identity management, Kubernetes server-side security.
Cloud, edge & more
FEDn enables FL on all kinds of servers and devices, including mobile.
Integrate anywhere
Uses industry standard technology and follows best practices for secure communication.
Usability
FEDn enables easy integration to existing MLOps. Or as a fast way to develop internal proof-of-concepts.
ML Framework Agnostic
FEDn is compatible with most existing and future machine learning frameworks.

FEDn SDK version 0.9 and FEDn Studio version 0.9 released!

The latest update enhances operational efficiency, robustness, flexibility, and user experience with guided setup, dedicated pages for models/sessions, better event filtering, and more.

FEDn SDK

The open-source SDK is engineered to prioritize scalability and security. It offers a seamless transition from development stages to large-scale deployments, consistently ensuring data security and effective model aggregation.  
Documentation »
SDK

FEDn SDK

FEDn, our open-source SDK, is designed for scalable and secure federated learning. It ensures a smooth transition from development to large-scale deployment, emphasizing data security and efficient model aggregation. Start with the resources and examples in our GitHub repository.

The FEDn Software Development Kit

Features

Model agnostic
Versatile machine learning model-agnostic framework.
Tiered architecture
Tiered architecture for efficient client coordination.
Enhanced security
Enhanced security with protected client ingress ports.
Docker and native
Support for both Docker-based and native deployments.
Monitoring
Comprehensive oversight through real-time event logging and training progress tracking.
Deployment Versatility
Flexible deployment options: all-in-one and distributed setups.

Get started

Begin with the SDK using our quick start guide, explore the API, or dive into the API reference.
Quick start guide
A quick start guide for starting a pseudo-distributed FEDn network using docker-compose.
Quick Start
Client API
The APIClient is a Python3 library that can be used to interact with the FEDn network programmatically.
Client API
API reference
Detailed documentation on the API, providing guidance for programmatic interaction with the FEDn network.
API Reference

FEDn Studio

Our cloud-native platform, delivers seamless and secure federated machine learning, designed to integrate effortlessly with your existing MLOps pipeline. It scales to any demand, ideal for both early-stage experimentation and fast iterations, as well as for on-premise deployment in production stages. Key Features below.
Documentation »
APP

FEDn Studio

FEDn Studio, our cloud-native platform, delivers seamless and secure federated machine learning, designed to integrate effortlessly with your existing MLOps pipeline. It scales to any demand, ideal for both early-stage experimentation and fast iterations, as well as for on-premise deployment in production stages.

Features

Security and encryption
Robust client authentication and secure encrypted communication.
Dashboards & visualizations
User-friendly visualizations and dashboards for real-time monitoring and training evaluation.
Deployment flexibility
Flexible SaaS and on-premises deployment options to meet various organizational needs.
Resource isolation
Project-specific resource isolation ensured through secure Kubernetes network policies.
Universal compatibility
Vendor-neutral compatibility, supporting bare-metal, public cloud, or on-premises setups.
Robust operations
Uses a Kubernetes infrastructure layer to enhance robustness and operational efficiency.

Get started

Access the FEDn Studio, explore the tutorial for a quick start, and register to get access.
FEDn Studio
Get access to FEDn Studio by following the link.
Sign Up
Tutorial
Follow the tutorial to get started with the FEDn Studio.
Quick start
Introduction
Learn more about the FEDn Studio in this introduction.
Learn
WorkshopGet StartedFrameworkUse caseBlogAboutTerms
Star us on GitHub
Star