Part 1: BBVA: Building a multi-region, multi-country global Data and ML Platform at scale

In an era where data drives financial innovation, BBVA, a global financial institution serving over 76 million customers across 25+ countries, partnered with AWS to revolutionize its data analytics capabilities. BBVA embarked on an ambitious three-year journey to transform its entire data analytics infrastructure. The Analytics, Data, AI (ADA) Global Data platform represents one of the largest and most complex cloud migrations in the banking sector, which marks a significant milestone in BBVA’s digital transformation strategy.

The scale of this transformation is remarkable. As of May 2025, the platform manages over 8.4 petabytes of active data and over 30,000 tables, while executing upwards of 100,000 daily processing jobs. This massive infrastructure serves a diverse user base of 6,500 advanced users, including data scientists and engineers, alongside 40,000 daily data consumers across seven countries on two continents. See Figure 1 for more usage statistics.

Figure 1- BBVA ADA Platform

BBVA successfully merged 16 separate tenants into one unified global platform, making this migration particularly noteworthy. Deployed across two AWS Regions serving BBVA’s major markets, the platform now processes over 1M analytics jobs monthly, supporting critical banking operations throughout Europe and the Americas. This consolidation not only streamlined operations but also improved efficiency and reduced operational complexity, setting new standards for financial services innovation. Recognized in 2025 with a Global Finance ‘The Innovators’ award as well as “The Banker’s Technology Awards 2025, Global” for Transformational Project.

Read the entire series:

• Part 1: BBVA: Building a multi-region, multi-country global Data Platform at scale
• Part 2: How BBVA processes a global data lake using AWS Glue and Amazon EMR
• Part 3: How BBVA built a global analytics and machine learning platform on AWS
• Part 4: Building a Multi-Layer Security Strategy for BBVA’s Global Data Platform on AWS
• Part 5: BBVA’s unified console: Key to successful global platform deployment
• Part 6: Effective sunset of the legacy data platform in BBVA: the migration methodology

The challenge

BBVA discovered that traditional solutions could not cope with its 40% data infrastructure growth year over year. BBVA faced this exact challenge as their data science teams pushed the limits of their on-premises systems. The bank’s data lake, crucial for storing and analyzing vast amounts of information, struggled with inefficient data partitioning. This led to longer processing times and difficulties in accessing important insights. Even with an advanced job scheduling engine, the rigid infrastructure didn’t scale to meet surging demands. The impact rippled across the organization, with daily data operations competing for limited resources.

• Regulatory reporting, like IFRS 9 or BCBS239 required massive data processing capabilities.
• Real-time liquidity calculations needed fast, reliable access to financial data.
• Machine learning models, vital for improving customer experiences, faced deployment delays.

The on-premises environment’s limitations in scaling and evolving BBVA’s data platform clearly demanded change. For a bank that won multiple awards for digital transformation, this technical bottleneck threatened its position as an industry innovator.

This is the story of how BBVA revolutionized its data infrastructure to maintain its position as a digital banking leader. From tackling immediate scaling challenges to building a foundation for future innovation, here’s what they learned along the way.

BBVA’s ADA Platform: Pioneering Cloud-Native Data & MLOPs Architecture

At the heart of BBVA’s digital transformation lies the Analytics, Data, and AI (ADA) Platform – a groundbreaking cloud-native architecture that reimagines how global banks manage, process, and derive insights from vast amounts of financial data. Multiple time zones created unique challenges for BBVA’s global operations. Each region had different business hours, peak processing periods, and specific operational requirements. This geographical distribution demanded sophisticated scheduling and orchestration mechanisms to ensure smooth operations across all regions. Built on AWS, this architecture sets a new standard for scalability, security, and innovation in the financial sector. BBVA designed the Global Data Platform following AWS Modern Data Analytics Reference Architecture deployed in two regions, one tenant on each continent as depicted below in Figure 2.

Figure 2- BBVA ADA Platform Global Deployment

A Multi-Account Strategy for Enhanced Security and Flexibility

BBVA’s ADA platform is more than just a data repository; it’s a comprehensive ecosystem for financial data processing and analytics where:

• Batch processing leverages the power of Amazon EMR for large-scale data transformations, enabling BBVA to process millions of transactions efficiently.
• BBVA develops and deploys AI models that enhance customer experience and optimize their operations, leveraging machine learning and data processing fully integrated capabilities, powered by Amazon SageMaker.

BBVA has established an impressive analytics ecosystem that combines robust governance with flexibility for innovation. At its core lies a data governance framework that handles security and lifecycle management while empowering teams to experiment freely. BBVA’s ADA platform employs a sophisticated multi-account strategy that goes beyond cloud implementations. This approach not only enhances security but also provides the flexibility needed to support diverse workloads across the bank’s global operations. Multiregion deployments of ADA are key to delivering the platform where BBVA operates to reduce latency for global operations (Data ingestion, reporting, etc.) using AWS CloudFormation StackSets automation. Figure 3 depicts BBVA ADA global data platform based on AWS modern data architecture combining the concept of structural accounts and business domain sandbox accounts for consumption.

Figure 3 – BBVA AWS Modern Data Architecture

The Data Nucleus: Centralized and Secure Data Account

At the core of the ADA platform is the Data Account, serving as the central nervous system for data operations:

• Amazon Simple Storage Service (S3) forms the foundation of BBVA’s data lake, housing petabytes of financial data with intelligent tiering for optimal performance and cost-efficiency.
• AWS Glue Data Catalog acts as a comprehensive metadata repository, offering a unified view of BBVA’s vast data landscape.
• AWS Lake Formation actively provides fine-grained access controls, ensures data security and controls data sharing across the organization.

To simplify interactions with these powerful AWS services, BBVA developed a custom API layer to comply with stringent security and performance requirements. This layer, based on a proper physical data modeling, accesses the data, delivering the expected information to the consumers across the entire BBVA organization. This innovation abstracts the management of both AWS Glue Data Catalog and AWS Lake Formation, letting BBVA’s data scientists and analysts focus on generating insights rather than managing infrastructure. For more information, please read more details on “Part 2: How BBVA processes a global data lake using AWS Glue and Amazon EMR”.

Powering Analytics: The Processing Powerhouse

The Processing Account is where the data transformation happens using Amazon EMR, fully integrated with the AWS Glue Data Catalog and AWS Lake Formation.

• Amazon EMR clusters execute complex Spark-based workloads, handling both production processes and experimental sandbox environments.
• Seamless integration with the central Data Catalog ensures that processing jobs always have access to the latest data and metadata.

Having Amazon EMR compute decoupled from the Amazon S3 storage layer fully integrated with Amazon Sagemaker AI permits BBVA to scale based on business needs. This provides the power for processing sophisticated financial analytics, leveraging massive batch capabilities while running in parallel with end-users processing engines and AI models, maintaining strict control over resource allocation and costs without disrupting data teams’ daily work.

Fostering Innovation: Sandbox Environments

Think of BBVA’s sandbox environment as a secure playground for data professionals. Each business unit has its own dedicated space where data scientists and engineers:

• Discover and prepare data.
• Test machine learning algorithms.
• Develop and refine models.
• Collaborate through code repositories like Bitbucket.

Data scientists and data engineers leverage Amazon Sagemaker Studio with its state-of-the-art IDE for developing machine learning models. In addition, AWS Glue supports interactive ETL sessions, enabling rapid experimentation and data preparation. Finally, Amazon Athena lets analysts run ad hoc queries directly against the data lake, promoting data-driven decision making across the organization.

BBVA designed these sandboxes to balance freedom and control to innovate while maintaining the security and governance required in the financial sector.

Security and Governance: The Bedrock of Trust

In the world of global finance, security isn’t just a feature – it’s a fundamental requirement. BBVA’s ADA platform sets new standards for data protection and governance:

• AWS Identity and Access Management (IAM), integrated with BBVA’s internal identity systems, ensures that only authorized personnel can access sensitive data and systems.
• AWS Lake Formation provides granular access control at the data level, allowing BBVA to implement the principle of least privilege across its data assets.
• Amazon AppStream 2.0 adds an extra layer of security, preventing data exfiltration and ensuring that sensitive information never leaves BBVA’s controlled environment.
• AWS CloudTrail meticulously logs every data access, providing a comprehensive audit trail for compliance and security analysis.

More information regarding ADA Security on “Part 4: Building a Multi-Layer Security Strategy for BBVA’s Global Data Platform on AWS.”

Streamlined Access Management

The BBVA ADA console reduces data subset access time from weeks to minutes, improving time to market for the company. Also improves the deployment time for new sandboxes depending on business needs, solving governance and security issues seamlessly. This revolutionary automation workflow helped BBVA keep the pace of innovation while maintaining strict security protocols. More information on BBVA ADA console is available in “Part 5: BBVA’s unified console: Key to successful global platform deployment.”

BBVA’s Cloud Migration Journey: Lessons from Moving a Global Bank to AWS

The Great Migration of 2023-2025

In 2023, BBVA embarked on an ambitious journey: moving its entire data analytics infrastructure to AWS. This wasn’t just another IT project – it was a transformational initiative that reshaped how one of the world’s largest banks handles data. BBVA identified resource management as a critical concern during the migration. Also, BBVA needed to address diverse skill sets across teams, implemented comprehensive training programs for new technologies, and managed change across multiple organizations. BBVA prioritized knowledge transfer and documentation to maintain operational continuity and maintain service quality throughout the migration process.

Setting the Stage

Before the big move, BBVA took a smart approach: cleanup first, migrate second. This “housekeeping” phase proved invaluable, reducing the migration scope by an impressive 40% as well as downsizing the migration effort.

Technical Innovation at Scale

BBVA created a robust catalog synchronization system to ensure catalog replicas will run at scale. The solution processed over 30,000 tables, handled tables with up to 2,500 columns, managed over 100,000 partitions and completed initial loads in just 15 minutes. This approach used a sophisticated architecture that involved AWS tools like AWS Step Functions, Amazon Simple Queue Service (SQS), and AWS Lambda. More information is available in “Part 6: Effective sunset of the legacy data platform in BBVA: the migration methodology”.

Five critical lessons learned and a strategic decision: Build vs Run

BBVA created a robust catalog synchronization system to ensure catalog replicas will run at scale. The success of the ADA global data platform hinged on a strategic decision: establishing a specialized team focused solely on data platform design and architecture build, distinct from operational teams. This separation of “build” versus “run” responsibilities ensured both innovation in design and excellence in execution. Figure 4 depicts the program phases as well as key lessons learned by the BBVA team.

Figure 4 – BBVA BBVA ADA Platform Migration Program

The team that designed and built the global data platform learned the following critical lessons during the migration journey as depicted in the abovementioned picture:

1. Hidden Complexity: Building the tools to fully automate the migration from the old platform presented the real challenge, rather than building the new platform itself.
2. Parallel Operations are Tough: The team learned that shorter parallel phases are better as BBVA faced challenges with end users that had to work managing two platforms simultaneously, when running over 50,000 daily jobs to effectively reach a success rate to give the “green” light for the “go-live” on AWS ADA platform.
3. Pre-Migration Cleanup is Essential: The ‘housekeeping’ phase reduced the migration scope by 40%, which made the entire project more manageable and efficient.
4. Regulatory Hurdles are Real: Multiple countries’ regulatory bodies posed greater challenges than BBVA anticipated. AWS Compliance teams provided crucial support throughout the process.
5. Plan for Shutdown from Day One: The team learned that a platform shutdown strategy needs to be baked into the project and business case from the start – not treated as an afterthought.

This migration represents more than just a technical achievement – it’s a testament to careful planning, regulatory diligence, and technical excellence. As banks worldwide consider similar moves, BBVA’s experience offers valuable lessons for successful cloud migration at scale.

BBVA’s Innovative FinDataOps: Pioneering Cost Control and Visibility

In response to the unique challenges of managing costs and optimizing resource utilization in a cloud-based financial analytics platform, BBVA has developed a groundbreaking FinDataOps practice. This innovative approach combines the principles of FinOps with the specific needs of data-intensive operations, creating a comprehensive framework for cost control and visibility across the ADA platform.

To prevent unintended cost escalations, BBVA designed and implemented a robust system of guardrails. This proactive approach includes:

• Custom-built guardrails deployed across sandbox environments.
• Real-time monitoring and automated shutdown of idle or over-provisioned resources.
• Integration with CI/CD pipelines to enforce cost-optimized configurations in all deployments.

BBVA has deployed structural guardrails, which apply to the data processing accounts as depicted in the ADA Global architecture. BBVA was aware based on experience that where “humans” operate is where there is greater risk of unintended consumption. Therefore, by default within the sandbox accounts is where BBVA deployed the majority of guardrails and controls.

Figure 5 – BBVA ADA Guardrails for unintended consumption: FindataOps

The FinDataOps solution depicted in figure 5 leverages the power of AWS Cost and Usage Reports along with a combination of detective and preventive guardrails. The FinDataOps team deployed these guardrails in areas where human interaction is frequent, particularly across sandbox accounts where data scientists and analysts conduct their work. By implementing this comprehensive FinDataOps practice, BBVA achieved remarkable results:

• Even though storage and end users’ growth was over 40%, BBVA leveraged the abovementioned analytics guardrails to improve resource utilization. BBVA reduced overall cloud spend by 28% within the first six months after the “go live” of the platform in Europe.
• Sandbox owners actively manage end users’ activities, promoting cost awareness among data scientists and analysts.
• BBVA reduced analytical environment preparation time by 94%.
• Improved time-to-market for new analytics projects attributed to streamlined resource allocation.

BBVA’s FinDataOps practice represents a significant step forward in the management of cloud costs for financial analytics. By combining rigorous financial governance with the agility required for cutting-edge data science, BBVA has created a model that balances innovation with fiscal responsibility. This approach not only optimizes costs but also enhances the overall value derived from cloud investments, positioning BBVA at the forefront of efficient, scalable, and responsible use of cloud technologies in the financial sector.

For a more detailed exploration of BBVA’s FinDataOps implementation, including technical specifications and case studies, please refer to “Part 3 – How BBVA built a global analytics and machine learning platform on AWS” in this series.

The ADA successful roll out Story

By late 2024, BBVA achieved a major milestone: completing the european migration. The process included running parallel solutions on premise and AWS (“dual run”) to ensure Batch generated the expected output as well as end users were able to operate the platform accordingly. On May 16th, 2025, BBVA took a significant step forward for its operations in America and its data-driven transformation strategy, with the successful implementation of BBVA’s global ADA data platform in Mexico and Colombia.

Conclusions: BBVA’s Cloud Transformation: Redefining Global Banking with AWS

BBVA’s journey to AWS shows how a global financial institution can successfully transform its data analytics capabilities through cloud adoption. BBVA leveraged AWS services to transition from 16 separate tenants to a single, cohesive Global Data Platform that handles petabytes of active data, 30,000 distinct datasets, Support for 6,500 advanced users and 40,000 data consumers across seven countries. Key achievements include:

• 40% reduction in operational costs.
• 10x faster model development and training.
• 94% reduction in environment setup time.
• Enhanced business recovery services and operational resilience while eliminating the prohibitive costs of on-premises redundancy.
• Enterprise-grade security by leveraging AWS IAM, AWS Key Management Service (KMS) and AWS CloudTrail, providing comprehensive data protection, traceability, and access management.
• Rapid innovation of AWS capabilities for analytics and machine learning while optimizing resources through FinDataOps practices that will be extended to leverage Generative AI as well soon.

This transformation succeeds beyond technical metrics as BBVA positions itself as an attractive destination for top tech talent by adopting AWS technologies. The opportunity to work with cloud technologies and contribute to large-scale data analytics projects has significantly boosted BBVA’s appeal in the competitive tech job market.

As BBVA expands its cloud footprint to Peru and Argentina in late 2025, their experience provides a valuable blueprint for other financial institutions considering large-scale cloud migration. BBVA roadmap for ADA will add capabilities to manage unstructured data seamlessly integrating advanced language models and Gen AI, leveraging AWS services to process data in real-time, making instant decisions and detecting anomalies while streamlining data science workflows, from data exploration to model deployment to maximize asset reuse and accelerate project delivery reducing development time. BBVA proves that organizations can not only achieve full cloud migration but benefit significantly from it, earning recognitions like “The Banker’s” Technology Awards 2025.