Skip to main content

Data Owner: Setting Up Your Datasite

Work in Progress

This guide is being updated for syft-flwr data owners. Check back soon!

Overview​

As a data owner, you have two main ways to set up and manage your datasite for participating in federated learning experiments with syft-flwr.

The RDS Dashboard provides a user-friendly web interface for managing your datasite and participating in federated learning jobs.

The dashboard allows you to:

  • Set up your SyftBox datasite through a visual interface
  • Browse and review incoming FL job proposals
  • Approve or reject jobs with a single click
  • Monitor running FL clients and training progress
  • Manage your datasets and privacy settings

The RDS Dashboard is ideal if you prefer a graphical interface and want to get started quickly without writing code.

Location: The dashboard is available in the rds-dashboard directory alongside this documentation project.

Option 2: Using Code (syft-rds API)​

For more advanced users or automated workflows, you can programmatically manage your datasite using the syft-rds API.

The syft-rds Python library provides:

  • Programmatic control over datasite setup and configuration
  • API-based job submission and approval
  • Scriptable dataset management
  • Integration with your existing data pipelines
  • Automation of repetitive tasks

This approach is ideal for:

  • Power users who prefer command-line interfaces
  • Automated deployment scenarios
  • Integration with existing data infrastructure
  • Custom workflows and batch operations

Documentation: Detailed API documentation coming soon.

Getting Started​

Prerequisites​

Before setting up your datasite, ensure you have:

  • SyftBox client installed on your machine
  • Python 3.11+ (for programmatic approach)
  • Network connectivity to the SyftBox network
  • Datasets you want to make available for federated learning (optional)

Next Steps​

Choose your preferred setup method:

  • New to federated learning? → Start with the RDS Dashboard
  • Comfortable with code? → Use the syft-rds API for more control

After setup, you'll be able to:

  1. Configure your datasite and make it discoverable
  2. Review and approve FL job proposals from data scientists
  3. Run FL clients to participate in privacy-preserving model training
  4. Monitor your contributions to federated learning experiments

Detailed step-by-step guides for both approaches coming soon!