Real-Time Data Flows
Erik shares insights on how Spotify handled real-time data flows for music recommendations, using cron jobs and workflow schedulers. They discuss the challenges of dealing with new tracks and users, as well as the complexities of generating training data sets from real-time features. Michael highlights the importance of hiding this complexity and enabling it within the modern data stack for various use cases, such as building a production fraud detection system.In this clip
From this podcast
MLOps.community
The Future of ML and Data Platforms // Michael Del Balso - Erik Bernhardsson // Coffee Sessions #57
Related Questions