ART-IE Federated Machine Learning
AI & Data
Semester programme:Complex Software Systems
Client company:ART-IE
Project group members:Renato da Silva Pereira
Robert Figaroa
Heinrich Cornelis
Denzel Huijbers
Himal Aryal
Project description
How can a federated learning platform be designed and implemented that enables collaborative AI model training across multiple clients while remaining compliant with GDPR?
Context
The project is a federated learning platform that enables multiple clients to train AI models locally and submit only their model updates to a central service, addressing the growing need for privacy-preserving machine learning in a GDPR-regulated environment.
Results
The main deliverable is a federated learning platform built across multiple microservices. The key development this semester was migrating from a user-based to a project-based model structure, enabling users to create projects, invite collaborators, and collectively contribute trained model updates. This shift unlocks real federated learning workflows. Validated through functional and integration testing with high code coverage, the platform sits at TRL 4, demonstrated and validated in a lab environment.
About the project group
Our group consists of five 6th-semester ICT students (Complex Software Systems) at Fontys Eindhoven. We worked full-time, four days a week, in two-week sprints using a GitLab issues board to track our work.