Using Feathr in an Azure Machine Learning Environment

Feathr is the feature store that is used in production in LinkedIn for many years and was open sourced in April 2022.

Read the announcement on Open Sourcing Feathr and Feathr on Azure.

In this repo we use the official Feathr repo to demonstrate the use of Feathr in an integrated Azure Machine Learning Environment. We will use some different sample datasets to practice the concepts as well. Feel free to contribute with us 😉.

We use the same high-level architecture proposed on Feathr documentation

Getting Started

You can use the same Quick Start from Feathr to deploy the resources in your own Azure's subscription.

• Open your Azure ML environment and create a Computer Instance. The easiest way is to use the own UI to create it.

• Next, open the terminal (or a notebook) and install Feathr client on the created compute instance

pip install -U feathr

After restarting the Azure ML Kernel, you should be able to import Feathr:

from feathr import FeathrClient

Running Feathr on Azure ML

Check our sample notebook that demonstrate how to use Azure ML to train the model and Azure Databricks as the spark engine. In the end we'll have a realtime inference endpoint deployed ... 😍