AT&T and H2O.ai built an artificial intelligence (AI) feature store to manage and reuse data and machine learning engineering capabilities. The AI Feature Store distributes the features data scientists, developers and engineers need to build AI models.
The AI Feature Store is in production at AT&T, meeting the high performance, reliability and scalability required to meet AT&T’s demand. AT&T and H2O.ai announced that the same solution in production at AT&T, including all its industry-first capabilities, will now be available as the “H2O.ai Feature Store” to any company or organisation.
Data scientists and AI experts use data engineering tools to create features, which combine relevant data and derived data that predict an outcome (churn, likely to buy, demand forecasting).
Building features is time-consuming work, and typically data scientists build features from scratch every time they start a new project. Data scientists and AI experts spend up to 80 per cent of their time on feature engineering, and because teams do not have a way to share this work, the same result is repeated by teams throughout the organisation.
Also Read: H2O.ai Data Scientist #1 in Kaggle Rankings
Also, features must be available for both training and real-time inference to avoid training-serving skew, which causes model performance problems and contributes to project failure. Feature stores allow data scientists to build more accurate features and deploy these features in production in hours instead of months.
Until now, there weren’t places to store and access features from previous projects. As data and AI are and will continue to be necessary to every business, demand is growing to make these features reusable. Feature stores are a critical component of the infrastructure stack for machine learning because they solve the most complex problem with operationalising machine learning — building and serving machine learning data to production.
AT&T carries more than 465 petabytes of data traffic across its global network on an average day. When you add in the data generated internally from our different applications, in our stores, among our field technicians, and across other parts of our business, turning data into actionable intelligence as quickly as possible is vital to our success. AT&T’s implementation of the AI Feature Store has been instrumental in helping turn this massive trove of data into actionable intelligence.
The H2O.ai Feature Store includes integration with multiple data and machine learning pipelines, which can be applied to an on-premise data lake or leveraging cloud and SaaS providers.
The H2O.ai Feature Store also includes Automatic Feature Recommendations, letting data scientists select the features they want to update and improve and receiving recommendations. The H2O.ai Feature Store recommends new features and feature updates to improve the AI model performance.
Also Read: Which Tech Trends Will Impact Your Business?
