This repository hosts the codebase for a low-code Federated Learning platform designed for training and evaluating radiology AI models on locked data silos — enabling collaboration without data sharing.
Built on NVIDIA FLARE (NVFlare), this platform facilitates federated AI development while ensuring privacy and compliance in sensitive medical environments.
⚠️ Prerequisite:
To use or extend this platform, basic familiarity and hands-on experience with NVFlare is required.
The repository is organized into three main components:
| Module | Description |
|---|---|
admin-backend |
A FastAPI server with PostgreSQL, serving as the interface to NVFlare's Admin Server. |
master-backend |
A FastAPI server that connects with NVFlare's Master Server for orchestrating FL tasks. |
frontend |
A React-based low-code interface that enables users to configure, train, and monitor FL jobs easily. |
- ✅ Low-code interface for federated AI development
- 🔒 Data never leaves the institution (compliant with data privacy laws)
- ⚡ Fast deployment using Docker (under 1 hour on AWS)
- 📈 Integration with NVFlare for robust federated learning workflows
- 🧪 Seamless training/testing workflow for radiology models
The entire platform is containerized and can be deployed on AWS in under 1 hour.
🛠️ Want to deploy this in your own cloud infrastructure?
Contact us for deployment support and custom integration.
Have questions or want to contribute? We'd love to hear from you!