With more than 13 million members, United Services Automobile Association (USAA) provides insurance and financial services to military families. Owned by its members, USAA's sole focus is serving those members, and AI helps it to improve its products and services.
As data scientists experimented iteratively with new AI models, they made new data sets. When a data set is used for training an AI model, it must be kept for regulatory purposes. However, tracking what was in use and what could be wiped became a challenge, so the immense amount of data grew and grew.
With Domino as its in-house AI platform, USAA built a new solution architecture that increases the governance for experimental data sets. It delivers:
USAA is looking at Domino’s feature store to cut the time that data scientists spend transforming data.
The new system provides governed and audited access to data for experimental AI training. Using it has enabled USAA to stop data from growing, helping to reduce costs while giving AI developers flexible access to the data they need.
for AI experimentation
and deletion of sandboxes
for building models
With Domino, we can seamlessly connect [to our experimental databases] just the way we would to any other database.
— Cody Nolden, Principal AI Architect, USAA
Source: Breaking barriers: How USAA is redefining data access & governance in financial services session delivered at RevX New York.
Domino Data Lab, Inc. Made in San Francisco.
Visit Domino Data Lab homepage