Feast

Overview

Feast is an open-source feature store for machine learning, written in Python. It centralizes the storage, discovery, and serving of features for both training and online inference workflows. By providing a consistent feature engineering and serving layer, Feast helps teams avoid duplication and ensure feature correctness across models.

Best for

Best for
Teams building ML pipelines who need a standardized, open-source feature store to manage and serve features consistently

Use cases

• Serving historical features for model training from a centralized repository
• Pushing and serving real-time features for online model inference
• Managing feature definitions, metadata, and lineage across multiple ML projects

Notes

Feast is an open-source feature store for machine learning, written in Python. It centralizes the storage, discovery, and serving of features for both training and online inference workflows. By providing a consistent feature engineering and serving layer, Feast helps teams avoid duplication and ensure feature correctness across models.

7,063 stars on GitHub. Last updated 2026-06-01. Licensed Apache-2.0.

Use cases

• Serving historical features for model training from a centralized repository
• Pushing and serving real-time features for online model inference
• Managing feature definitions, metadata, and lineage across multiple ML projects

Pros

• Open source with strong community support (7k+ GitHub stars)
• Provides a unified API for both batch and online feature serving
• Integrates with common data stores like BigQuery, Snowflake, and Redis

Cons

• Operational overhead: requires maintaining separate infrastructure (e.g., online store, registry)
• Limited built-in feature engineering capabilities compared to some proprietary alternatives
• Maturity and stability may not match enterprise-grade managed feature stores

Indexed from awesome-llmops and enriched against its public facts.