Federated learning has emerged as a popular alternative to traditional methods of training models to combat privacy issues and facilitate cooperation across geographically dispersed devices and servers. This chapter examines the technical foundations of federated learning, presenting a thorough comprehension of the decentralized learning paradigm and its potential to address data ownership and privacy concerns. It draws attention to the transition from centralized model training to decentralized federated learning, where data is stored locally on users’ devices yet shared for the sake of learning. This chapter explores the fundamental aspects of Federated learning-based Recommendation System. It also explores the various client devices including smartphones, tablets, and IoT devices, and their consequences for federated learning. Furthermore, the chapter explores the server aggregator, responsible for coordinating model updates from client devices and aggregating them to form a global model. It addresses the challenges related to communication, synchronization, and security in the context of server aggregation. The learning algorithms deployed for federated learning are also discussed. This chapter introduces federated optimization techniques and their privacy-preserving variations, including Federated Averaging and Federated Stochastic Gradient Descent and focuses on how these algorithms promote privacy-preserving collaborative recommendation system. In conclusion, this book chapter gives an in-depth analysis of the technical fundamentals of federated learning. It highlights the potential of federated learning to solve privacy issues, make decentralized training more scalable, and give users greater influence over their data.