VTB has replaced its own big data analytical platform

2023: Transition to Arenadata Hadoop

VTB Bank implemented a project to migrate its own big data analysis and processing platform to the Russian Arenadata Hadoop solution. Now the entire process of preparing data for modeling with the subsequent calculation of artificial intelligence (AI) models is carried out on an import-independent platform. Computing power is more than doubled, greatly speeding up data processing. The total amount of platform data exceeded six petabytes. The bank announced this on December 15, 2023.

The data needed to build models of artificial intelligence, analysis, reporting and advanced analytics are available to business users from more than 200 source information systems. They are stored in a highly reliable geodistributed disaster recovery architecture. The domestic solution provides support for key business processes of management and regulatory reporting, as well as the development of advanced analytics models. More than 300 AI models are already running on the platform: customer scoring models, models for CRM tasks, models for assessing customer income, anti-fraud models and many others.

The value of the data required for analytics and artificial intelligence models is in their relevance, completeness and correctness. It was these characteristics, as well as modern data processing tools, that made it possible to analyze data much faster, build artificial intelligence models and use their results for the benefit of VTB customers, "said Maxim Konovalikhin, Head of Data Analysis and Modeling, Senior Vice President of VTB.

Fast processing, reliable data storage and wide functionality are among the key elements of the platform for analytics and building artificial intelligence models. Thanks to the implemented project, the bank managed to fulfill one of the key tasks in ensuring technological sovereignty, - said Nikita Rybchenko, head of the department for technological development of general bank systems, senior vice president of VTB.