databricks-feature-engineering

What it is and what it does

The Databricks Feature Engineering client is a Python library for managing feature tables within the Databricks ecosystem. It provides APIs to create, read, and write feature tables, train models using feature data, and publish tables to online stores for real-time serving. The library also supports upgrading workspace feature table metadata to Unity Catalog.

This is a Databricks-native tool that runs on Databricks runtimes and integrates with the broader Databricks ML platform. It depends on mlflow-skinny for experiment tracking, boto3 and azure-cosmos for cloud storage backends, databricks-sdk for platform communication, and standard ML libraries like numpy and protobuf. The library is actively maintained and has no known security vulnerabilities.

Use it for:

• Build and manage feature tables in Databricks for ML model training pipelines.
• Publish computed features to online stores for low-latency real-time inference.
• Migrate workspace-level feature tables to Unity Catalog for governance and access control.
• Integrate feature engineering workflows with MLflow experiment tracking in Databricks.
• Unit test feature engineering logic locally before deploying to Databricks clusters.

Worth the install?

AI-flagged interpretation of the facts on this page — verify before relying

Databricks Feature Engineering client for creating, managing, and serving feature tables within Databricks workspaces, including training models on feature data and publishing to online stores.

Yes, if you are working within Databricks and need to manage feature tables or integrate feature engineering into ML workflows. The library is actively maintained, has low install friction, and integrates with standard Databricks ML tools. However, the proprietary license restricts use to Databricks Platform Services; confirm your Databricks agreement permits this library before production deployment.

Install

databricks-feature-engineering on PyPI

pip install databricks-feature-engineering

uv add databricks-feature-engineering

poetry add databricks-feature-engineering

Installing databricks-feature-engineering

Before you install

Low install friction with a pure-Python wheel. Active maintenance with a release 25 days ago. Depends on mlflow-skinny, boto3, azure-cosmos, databricks-sdk, and other common libraries, all standard in ML environments.

License in practice

Licensed under Databricks Proprietary License with unclear treatment. Use is restricted to connection with Databricks Platform Services under a separate agreement; the license terminates if that agreement ends or is breached. Not open-source; review Databricks terms before deployment.

Quickstart

pip install databricks-feature-engineering

from databricks.feature_engineering import FeatureEngineeringClient

client = FeatureEngineeringClient()

Requires Python >=3.8.10 and can only run on Databricks runtimes; local/CI use is limited to unit testing.

Verify before relying

• Specific feature table schema requirements and supported data types.
• Online store provider support and configuration details beyond AWS/Azure/GCP.
• Performance characteristics for large-scale feature table operations.
• Backward compatibility guarantees across minor versions.

Package facts

Evidence: databricks_feature_engineering-0.16.1-py3-none-any.whl

Keywords: databricks, feature engineering, feature store

Tags

More Artificial Intelligence packages