Feature store
A system for defining, storing and serving model features consistently for both training and live predictions. It lets teams reuse features across models and ensures production uses the same logic as development.
Why it matters
When each model team calculates features separately, definitions drift and the same work is repeated. Differences between training and serving logic can cause models to behave differently in production from how they were tested.
A feature store is not always necessary. For organisations with a small number of models, well-governed feature tables in the warehouse may be enough; the principle of consistent, versioned feature logic matters more than a specific product.
In practice
For example, a UK lender running credit, fraud and collections models might hold shared features, such as recent missed payments, in a feature store so all three models use the same definition and point-in-time history.
Where Rodan fits
Rodan designs feature management proportionate to the number and criticality of models in AI and Decision Systems engagements.