Zalando is Europe’s largest online fashion platform, operating in over 25 countries and serving over 50 million customers. It uses the Hopsworks feature store to support real-time AI use cases, such as personalized recommendations. In this talk, we will look at systems aspects of operating a feature store at scale. In particular, we will explore challenges such as:

• High availability through replicated data across availability zones.
• Resilience by federating online feature stores to reduce the blast radius
• The pros and cons of a Kubernetes (4.x) environment over an EC2 deployment (3.9), and why this migration to Kubernetes was necessary.
• Auto-scaling to optimize resource utilization for daily and seasonal workload changes.
• Implementation of Infrastructure as Code utilizing the Hopsworks Managed Clusters API.