← All projects

Featureform

The Open-Source Virtual Feature Store

Data & Analyticsfeature-storemachine-learningmlopsopen-sourcedata-infrastructurepythonreal-time
Featureform screenshot

About

Featureform is an open-source virtual feature store that transforms existing data infrastructure into a machine learning feature platform. It allows data scientists and ML engineers to define, manage, version, and serve features, labels, and training sets using a declarative Python API on top of tools like Databricks and Snowflake. The platform also provides monitoring, governance, and collaboration capabilities for ML teams.

Problem

ML teams struggle to consistently define, version, share, and serve features across training and inference pipelines without duplicating work or losing data lineage.

For

ML engineers and data scientists building and managing machine learning features at scale

How it works

Users define features, transformations, and training sets via a Python API, and Featureform orchestrates the pipelines across existing infrastructure like Snowflake and Databricks, serving features for both batch and real-time inference.

Business model

open-source

Status

launched

Similar projects