openskills.info
Course catalog
Feature Stores with Feast logoCourse Preview

Feature Stores with Feast

A feature store organizes and serves the input data that machine-learning models use. Feast connects existing batch and online data systems so the same named features can support historical training and live inference.

itArtificial intelligence and machine learning

Feature Stores with Feast

A feature is a value supplied to a machine-learning model, such as a driver's completed trips in the last hour. A feature store gives those values shared definitions and retrieval paths. Feast is an open-source feature store that connects existing data infrastructure to model training and inference.

The central problem is time. Training needs historical values as they were known at each labeled event. Online inference usually needs the latest values for one or more entity keys. Building those paths separately can produce training-serving skew: the model trains on values created or joined differently from the values it receives in production.

Architecture and data flow

Feast separates feature definitions from feature data. A feature repository contains Python definitions and a feature_store.yaml configuration. Running feast apply validates those definitions, updates the registry, and may configure provider resources.

The main objects are:

Continue the course

This section is part of the paid course.

See pricing to subscribe, or log in if you already have access.

Where this skill leads

Relevant careers

See how this topic contributes to broader role-level skill maps.

Sources