Federated Learning as a Distributed Systems Problem: Designing Production-Grade ML Systems in Python.

Traditional machine learning pipelines break down under privacy constraints, data silos, and the realities of distributed systems at scale. In this talk, you will learn how to design production-grade federated learning systems in Python that address these challenges in real-world environments. We will explore the core distributed systems patterns behind federated learning — including client orchestration, communication protocols, fault tolerance, secure aggregation, and observability — and show how these concerns are implemented using practical Python-based frameworks such as Flower, without losing focus on the underlying engineering principles that make these systems reliable, secure, and scalable