Date: May 20, 2025
Time: 3:00 PM – 4:00 PM ET
Join us on May 20, 2025, from 3 – 4 PM ET, for “Federated Learning with Microservices on FABRIC,” the next session in our Stitching Together Innovation with FABRIC Users Webinar Series. This session is designed for new and experienced researchers interested in leveraging FABRIC’s cutting-edge capabilities. The webinar will feature a 30-minute tutorial providing an accessible overview of how FABRIC can support ambitious experiments, including a showcase of Federated Learning with microservices on the FABRIC testbed by Alexandros Koufakis and Jake Jongejans.
Alexandros, Jake, and their colleagues have leveraged FABRIC to forecast energy consumption and predict airplane maintenance needs using federated learning. They have utilized FABRIC to train models for global applications and to manage larger volumes of data. Moreover, they optimized the federated learning use-case on FABRIC by automating deployments across multiple intercontinental sites to research into the effects of network topology on federated learning training.
The session will include a demo of the experiment and a discussion on how FABRIC can empower other researchers, including its robust support and low barriers to entry. The webinar will conclude with a 15-minute Q&A session to address audience questions and provide insights on using FABRIC for your experiments. Don’t miss this opportunity to discover how FABRIC can help advance your research. Register now to secure your spot!
Register Here
Presenters
Alexandros Koufakis
Alexandros is a Scientific Programmer at the Complex Cyber Infrastructure (https://cci-research.nl) group of the University of Amsterdam. He is involved in research on microservices, federated learning, and building energy consumption prediction. He is the principal developer of the DYNAMOS project (https://github.com/Jorrit05/DYNAMOS), a middleware for complex data sharing use cases developed by the University of Amsterdam.
Jake Jongejans
Jake Jongejans is researching the effect of geographical policies on the training of Vertical Federated Learning models, at the Complex Cyber Infrastructure research group at the University of Amsterdam. His research utilises DYNAMOS, a dynamic microservice architecture, that allows policies to influence where code is allowed to execute and data is allowed to go. Some of his interests are writing, ballroom dancing, playing piano and working with functional and typed languages.