FL.IN.NRW

Description

Industrial data is often distributed across multiple organizations and cannot be shared due to security, ownership and regulatory constraints. At the same time, companies increasingly rely on data-driven approaches to optimize processes and develop new industrial AI applications.

FL.IN.NRW addresses these challenges by leveraging federated learning to enable secure and decentralized data utilization. Instead of sharing raw data, machine learning models are trained collaboratively across partners while data remains locally stored. This approach allows companies to unlock the value of distributed data while maintaining full control and data sovereignty.

Within the project, gemineers contributes its expertise in Digital Twin Management and data integration, supporting the development of scalable and interoperable solutions for industrial AI applications across distributed data ecosystems.

Role of gemineers in FL.IN.NRW

Within the FederatedLearning.IN.NRW project, gemineers takes a central role in the conception, development, and integration of the data infrastructure and digital twin-based framework for decentralized AI model development.

Key responsibilities include:

• Leadership of Work Package 1, including coordination of partners, definition of use cases, and specification of requirements for data acquisition and digital infrastructure
• Design and implementation of the data acquisition system, including integration of sensor-based systems, connection of heterogeneous data sources, and setup of secure and structured data storage
• Development and provision of the digital twin infrastructure, enabling process-parallel data acquisition, preprocessing, and standardized data interfaces for downstream AI services
• Contribution to data preprocessing and ML pipeline development, including support for synchronization of heterogeneous data and deployment of preprocessing services
• Support in the integration and orchestration of AI models and FL services, including containerization, infrastructure provisioning, and orchestration tool deployment
• Extension and adaptation of the data infrastructure for Federated Learning, ensuring seamless integration of data sources with the FL platform

Overall, gemineers is responsible for establishing the data-centric and digital twin-based foundation of the project, enabling secure, decentralized, and scalable development of AI models using Federated Learning technologies.

Summary

FL.IN.NRW enables secure, cross-company use of industrial data through federated learning, allowing machine learning models to be trained collaboratively while data remains locally stored. The project addresses key challenges of data sovereignty, interoperability, and decentralized AI.

Kick-off of the FederatedLearning.IN.NRW project at Fraunhofer IPT, Aachen