“Feature Stores”, with their dreary, opaque moniker, might not seem like the sexiest subject.
But they’re an essential part of the AI systems that businesses — and consumers, for that matter — use every day. This is why they are attracting more and more attention and investment from venture capitalists, who see market opportunities developing in the distant future.
AI systems are made up of many components, one of which is functionality. Features are the individual variables that act as inputs into the system. When thinking about features, it can be helpful to visualize a table, where the data used by AI systems is organized into rows of examples (data from which the system learns to make predictions) and columns of attributes (data describing these examples). Features are attributes used to describe each instance – an AI spam detection tool might use features like words in the body of the email, for example, or a sender’s contact information.
Working with features tends to be an ad hoc process within a single AI system. But at the enterprise scale, where data science teams are responsible for maintaining tens to thousands of systems, a place to manage and track functionality becomes a necessity.
Enter the Feature Store, a centralized repository for organizing, storing, and serving the features that AI systems rely on. Introduced as a concept by Uber in 2017, Feature Stores provide a unified place to create and share features across different teams in an organization.
“Feature stores sit at the intersection of data and machine learning,” Michael Del Balso, CEO of Tecton.ai, a startup developing feature store software for enterprises, told TechCrunch in a post. E-mail. “[Feature stores are] an essential part of the “MLOps” stack, as they enable data teams to quickly and reliably build high-quality features using real-time data and serve those features in production for in-time inference real. They act as an interface between data and [AI] models.”
Beyond a simple database, feature stores allow data engineers to see statistics about features, including which features have been used, where they have been used, and the impact they have had on the models. Feature stores also transform data, allowing users to aggregate, filter, and join features without necessarily needing to code. (Consider grouping orders in a restaurant to get the value of the characteristic “number of orders in the last 30 minutes”.)
Del Balso explained, “Advanced feature stores… automate production pipelines to collect data from batch data sources and real-time sources, transform the data in real-time, and store the data in the offline store. online and online. [They often also] include built-in monitoring capabilities to monitor pipeline health, data drift, service levels and more.
Feature stores promise to improve collaboration between teams while streamlining the development of AI systems. As demand grows, tech giants and startups like Tecton are developing products to meet the needs – and investors are enthusiastically backing them.