Centralized repository for storing, managing, and serving machine learning features for consistent use across training and production. Like a shared ingredient pantry that ensures everyone uses the same quality ingredients.
A feature store ensures that the 'customer lifetime value' calculation used during model training is identical to what's used when making predictions.
AWS and GCP provide managed feature store products (SageMaker Feature Store and Vertex AI Feature Store) to create, store, and serve features consistently for training and online inference. Azure and OCI do not have a single, universally named first-party 'Feature Store' service; teams commonly implement feature stores using their ML platforms plus data stores (e.g., Azure ML/Databricks + Delta/SQL/Redis; OCI Data Science + Object Storage/Autonomous DB), or use open-source feature stores.