Customers Gain Access to Production-Grade, Business-Vetted Features Ready for Deployment, Powered by AtScale’s Semantic Layer

AtScale, the leading provider of semantic layer solutions for modern business intelligence and data science teams, announced that it has added support for the Feast feature store. Feast is an open-source feature store gaining rapid adoption in the enterprise data science community. Optimized to run on Snowflake’s Data Cloud, the AtScale-powered Feast feature store advances the adoption of applied artificial intelligence (AI) for the enterprise by granting users easy access to production-grade, business-vetted features that are ready for deployment.

Machine learning (ML) models are only as good as the data that fuels them. AI/ML projects are too often impeded or delayed due to inaccurate data or siloed data pipelines by users. Generating features is also a difficult and time-consuming task, and traditionally there has been no easy way to publish and locate previously defined sets of features. Deploying new data science initiatives remains slow and manual, even within the same team.

Marketing Technology News: Portada And Comscore Publish Their Report “E-commerce Marketing In Latin America: Data And Lessons…

By integrating AtScale’s semantic layer with a Feast feature store, organizations can ensure that there is a single, consistent and accurate source of data powering their central catalog of features. Teams can easily discover and re-use features already created by other data scientists – and can rest assured that those features are based on an accurate, real-time data source. By integrating closely with Snowflake, data science teams can expand their universe of features to include first party application data and third party data sourced from data providers. Furthermore, AtScale will write new features and predictions generated by ML models back into Snowflake for easy access by business analysts or data scientists.

“Today’s organizations simply don’t have time to waste re-inventing the wheel at every stage of the development process. That is the advantage of feature stores. Ensuring access to the previous knowledge and effort of data science teams speeds development time,” said Danny Chiao, Engineering Lead at Tecton / Feast. “By integrating AtScale’s semantic layer into the process, machine learning model features will be based on a single source of governed enterprise metrics and analysis dimensions.”

Feast incorporates a rapidly expanding set of capabilities and integrates with any ML platform to help companies improve their efficiency and effectiveness across the ML lifecycle. Feast is backed by Tecton, a Snowflake partner.

“As a Snowflake Premier Partner, AtScale has already proven its success in helping Snowflake customers build a semantic layer for data and analytics that can be accessed by any BI or AI tool,” said Miles Adkins, Senior Partner Sales Engineer, AI & ML at Snowflake. “They are the perfect fit to help our customers get the most out of their Feast feature stores.”

Marketing Technology News: MarTech Interview with Kirsten Allegri Williams, CMO at Optimizely

The benefits of incorporating AtScale’s semantic layer technology into the enterprise ML workflow include:

• Feature Creation: The creation of production-grade, business-vetted features from a no-code modeling and business intelligence (BI) metric store make development easier and faster (e.g., time-series modeling).
• Feature Serving: The process of publishing business-ready features is simplified, leveraging query virtualization techniques to any ML training pipelines.
• Feature Write-Back: Users can write and catalog important features in the semantic layer for discovery, impact reporting and lineage within their BI and AI tools.
• Prediction Write-Back: ML model predictions can be written and cataloged alongside features in the semantic layer for impact analysis, and for prediction vs. actual correlations within BI and AI tools.

“Organizations are desperately trying to move from experimental AI to applied AI, but lack business context and a mechanism to apply it at every stage of the machine learning lifecycle,” added Gaurav Rao, GM of AI/ML at AtScale. “This is what makes a business powered feature store for developers so exciting for the enterprise. The combination of AtScale, Feast, Snowflake and Tecton can help deliver the needed capabilities to power end-to-end machine learning pipelines for mission-critical business applications.”