This article analyzes modern methods for creating and optimizing intelligent recommendation systems based on data engineering in the media ecosystem. The increase in the volume of media content consumption, the growth of user behavior data, and the demand for real-time recommendations require the creation of a strong, stable, and efficient data engineering infrastructure for recommendation systems. The article covers media data flow management, recommendation model architecture, data processing technologies, hybrid recommendation approaches, and methods for scientifically optimizing system performance.