Palo Alto, 22nd April 2020 - We are pleased to introduce Hopsworks.ai, a managed cloud service providing Hopsworks and the Feature Store. Hopsworks is a platform for the design and operation of AI applications at scale. Hopsworks.ai allows you to effortlessly launch and manage Hopsworks clusters in your AWS account and integrate them with third party platforms such as Databricks and AWS SageMaker. For users of Databricks and AWS SageMaker, it enables your organisation to manage and share your machine learning data with the Feature Store and gives your Machine Learning teams the ability to develop, train and deploy AI applications at scale following best practices established by industry leaders such as Uber (Michelangelo) and Airbnb (Zipline).
Hopsworks.ai offers two product tiers, a free version and an enterprise version. The free version is for individuals or organisations that want to get started with Hopsworks and the Feature Store. The enterprise version provides advanced features and support to organisations building production machine learning applications at scale. The enterprise version is in early access and available upon request.
The free and enterprise versions have the following features, respectively:
Sign up for free. If you quickly want to try the platform without connecting an AWS account, then you can make use of 30 days of free demo access. All you have to do to go further and get your own cluster is to connect your AWS account, see Getting started with Hopsworks.ai.
If you are interested in the enterprise version of Hopsworks.ai, then reach out to us to apply for the early access program: Contact Logical Clocks.
For technical questions regarding Hopsworks.ai you can reach out to us on Hopsworks Community.
Hopsworks.ai seamlessly integrates with Amazon SageMaker and Databricks, providing them with a Feature Store, usable directly from notebooks in those platforms. The Hopsworks Feature Store also offers Python, Scala and Java libraries to support custom integrations. For detailed information, see Feature Store Integrations.
To connect from Databricks, we offer both a native PySpark/Spark SDK based on Dataframes and a Python SDK based on Pandas. The native PySpark/Spark SDK is recommended for production workloads but requires you to establish network connectivity by either setting up VPC peering or placing your Hopsworks cluster in the same VPC and availability zone as the Databricks cluster. See Databricks Quick Start for documentation for how to connect to the Feature Store from Databricks.
We provide a Python SDK based on Pandas that supports integration with Amazon SageMaker. See SageMaker Quick Start for documentation on how to connect to the Feature Store from Amazon SageMaker.
The Cloud, Python and Scala/Java SDKs can be used to build custom integrations. See Using the Feature Store from any Python environment (KubeFlow) for documentation on how to connect to the Feature Store. For any technical questions, you can reach out to us on Hopsworks Community.