Inside BBVA’s MLOps transformation: from data platform to scalable ML on AWS

This article describes BBVA's MLOps transformation of its ADA platform using Amazon SageMaker AI to scale machine learning capabilities across the global bank.

• Introduced ephemeral development workflows for isolated testing via pull requests, reducing development time 20-75%
• Implemented automated CI/CD pipelines with GitHub Actions and SageMaker Pipelines for consistent model deployment
• Built governance framework using SageMaker Model Registry with auto-approval and manual workflows enforcing separation of duties
• Centralized approval workflows using AWS Step Functions and EventBridge for real-time model lifecycle validation
• Achieved 40-55% cost reduction through automated resource cleanup and efficient pipeline versioning
• Integrated corporate validation including SonarQube, security scanning, and compliance checks into MLOps workflows

BBVA's MLOps transformation demonstrates how modernized ML practices deliver efficiency gains while maintaining governance and compliance standards required in regulated banking environments.