Data

When a company with millions of consumers such as DoorDash builds machine learning (ML) models, the amount of feature data can grow to billions of records with millions actively retrieved during model inference under low latency constraints. These challenges warrant a deeper look into selection and design of a feature store — the system responsible ...