TensorFlow Recommenders is a library for building recommender system models using TensorFlow. It provides tools for the full workflow of recommender system development including data preparation, model formulation, training, evaluation, and deployment. Built on Keras, it offers embedding layers and factorization models for predicting user preferences, such as for movie recommendations.