Case Study: Itaú achieves faster ML deployment and higher productivity with AWS Glue and Amazon SageMaker

How Itaú, Latin America’s largest bank, improved speed to market for ML models using Amazon SageMaker Studio

Itaú Unibanco, the largest private-sector bank in Brazil, faced a challenge with its on-premises machine learning infrastructure, which was slow, inflexible, and costly. The process to deploy ML models could take up to six months, creating a significant backlog and hindering the productivity of its data science team. To overcome this, the bank turned to vendor AWS and its Amazon SageMaker Studio service to modernize its ML operations on the cloud.

The solution implemented by AWS utilized a suite of services including AWS Glue for data integration and Amazon SageMaker Studio as the core development environment. This allowed Itaú to drastically reduce its ML model deployment time from six months to just 3-5 days, eliminating the waiting list entirely. The results included improved speed to market, increased productivity for over 3,200 users, significant cost savings, and greater standardization across its data science teams.

Itau

Diego Nogare

Ml Engineering Manager