Meet Rizemind#
Shared intelligence. Sovereign data.
Rizemind is an open-source framework for federated learning with verifiable multi-party coordination.
It is designed for developers and organizations that need to train or evaluate models across separate data environments without transferring the underlying datasets into a common repository.
Each participant trains locally. Raw training data remains in the participant’s environment. Authorized model updates are exchanged through the federated-learning process.
Rizemind adds participant authentication, cryptographically signed model updates, contribution measurement and ledger-backed records of selected training events. The available records and controls depend on the configured workflow. See model identity and records for how to interpret the model and version associated with an activity.
Arc deployment: Rizemind’s ledger-backed components are deployed on Arc Mainnet. See the deployment guide for the implementation, network configuration and verification steps.
Start here#
Built on Flower#
Rizemind builds on Flower. Flower provides the federated-learning foundation; Rizemind adds infrastructure for coordinating independent participants. Developers bring their own model, training procedure and evaluation metrics rather than adopt a proprietary Rizemind model.
See Integrating with Flower for the adoption workflow and the distinction between Flower examples and tested Rizemind support.
Choose your path#
Developers: start with an existing model and review the integration workflow.
Participating organizations: review the data and visibility boundaries before connecting a local participant to a federation.
Federation coordinators: review the roles and responsibilities, then select an appropriate example.
Further explanation#
Read the architecture for the workflow, capabilities and information boundaries. Explore illustrative use cases and the research foundation.