An open source feature store for machine learning

Feast is the bridge between your data and your machine learning models. It allows teams to register, ingest, serve, and monitor features in production.

Carbon (2)
  • Register

    Feast provides a registry through which to explore, develop, collaborate on, and publish new feature definitions. The registry is the central interface for all interactions with the feature store.

  • Ingest

    Easily ingest data from both batch and streaming sources into both online and offline feature stores, automating data management and making features available for serving.

  • Serve

    A feature retrieval interface that provides a consistent view of features in stores. Feast provides a point-in-time correct interface for training data, and a low-latency API for online serving.

  • Monitor

    Feast allows teams to confidently operate machine learning systems by publishing operational metrics, statistics, and logs to their existing production monitoring infrastructure.

    Architecture

    Architecture

    Teams contributing to Feast

    • Agoda
    • Gojek
    • Google Cloud
    • 1200px zulily logo 2019.svg
    • 0e2a0c6f 504a 4461 9ad3 9313db78bc62 1599663750816
    • Cimpress identity color retina

    Sign up for our newsletter

    Error. Your form has not been submittedEmoji
    This is what the server says:
    There must be an @ at the beginning.
    I will retry
    Reply