Evolution of Banking Data Platforms Over Two Decades
Over the past two decades, banking data platforms have undergone a significant transformation, evolving from traditional enterprise data warehouses to real-time, governed data and AI platforms. Samit Kumar Banerjee, an experienced professional in the banking industry, has witnessed this evolution firsthand through various engagements.
Banerjee's early work involved managing offshore delivery of data warehouse and analytics solutions for a major international bank. The landscape at the time included multiple country- and business unit-specific data warehouses, along with a global data warehouse that consolidated data across different regions and functions. The technology stack included IBM DB2, IBM DataStage, and COGNOS for analytics and reporting.
In later engagements, Banerjee shifted towards architecture and strategy, particularly during a diagnostic assessment for a prominent private-sector bank in India. This period saw the industry moving towards integrated and workload-optimised data warehouse systems. A subsequent engagement with a central bank in Southeast Asia exposed him to the broader functional purpose of banking data, which supports financial surveillance, economic analysis, and policy formulation.
Recent engagements have highlighted the evolution of banking data platforms. For a large public-sector bank in India, Banerjee was involved in modernising the data infrastructure using a hybrid-cloud model with technologies like Cloudera Data Platform and Red Hat OpenShift. Another engagement focused on developing an enterprise data lakehouse solution using IBM watsonx, which integrated advanced analytics, AI/ML use cases, and regulatory compliance.
Despite technological advancements, fundamental disciplines of data warehousing remain consistent. The layered architecture has evolved from raw data to integrated and business-ready information, supporting a wide range of consumers and use cases. Modern banking data platforms now require scalability, resilience, cyber security, and other non-functional capabilities to be integrated from the beginning.