A Feature Store for AWS Sagemaker and Databricks

October 11, 2019
in
Product Updates
by
Jim Dowling
Kim Hammar
Steffen Grohsschmiedt
Fabio Buso
Theofilos Kakantousis
Antonios Kouzoupis
Edited: First published 

Hopsworks Feature Store for AWS Sagemaker and Databricks

TLDR; At the tail end of 2018, we announced the world's first open-source Feature Store, as part of the Hopsworks platform. Now we are announcing support for the standalone Hopsworks Feature Store from both AWS Sagemaker and Databricks with IAM role-based integration.

Hopsworks Feature Store makes it easier for Enterprises to manage their machine learning features in a single place, share features across different teams, ensure consistent features between training and inference, and to perform time-travel queries for features (useful for generating new training data). "We want to make all the benefits of the Hopsworks Feature Store available to users on as many machine learning platforms as possible, such as AWS Sagemaker and Databricks. To date, many of our Enterprise Hopsworks customers have benefited tremendously from centrally managing and sharing their features with our Feature Store, and we have now extended its capabilties to make both the online and offline Feature Store accessible from 3rd party platforms", says Dr. Jim Dowling, CEO of Logical Clocks.

https://codepen.io/okawa-h/pen/abbVXQN.html

<script src="https://gist.github.com/omoindrot/dedc857cdc0e680dfb1be99762990c9c.js"></script>

Read more about the Hopsworks Feature Store.