Back to all tools
Feast
Open-source feature store — serve ML features consistently from training to production.
0Open Source
AI Data
Overview
Feast is an open-source feature store for machine learning that manages and serves features consistently between offline training and online serving, eliminating training-serving skew.
Key Features
- Unified feature definitions for offline and online serving
- Point-in-time joins for training dataset generation
- Low-latency online feature retrieval (Redis, DynamoDB)
- Feature versioning and registry
- Supports BigQuery, Snowflake, S3, and Spark
- SDK for Python and Go
Real-World Workflows
Eliminate training-serving skew
- 1Define feature views pointing to your data sources
- 2Generate training datasets with point-in-time correct joins
- 3Materialize features to online store (Redis) before serving
- 4Retrieve real-time features at inference with get_online_features()
Getting Started
pip install feast # Initialize a feature repo feast init my_feature_repo cd my_feature_repo # Apply feature definitions feast apply # Materialize features to online store feast materialize-incremental $(date -u +"%Y-%m-%dT%H:%M:%S")
Compare Alternatives
See how Feast stacks up against similar tools.