Managing the Mountains of Data

At the 2025 edition of the Technoviti Conclave & Awards, Banking Frontiers curated a high-stakes panel of thought leaders to address an increasingly critical imperative within the BFSI sector: the shift from data abundance to actionable intelligence. As the digital economy matures and customer interactions proliferate across channels, BFSI institutions are inundated with data, rich in potential but difficult to tame.

- 

ARCHITECTURE FOR INTELLIGENCE

Building on this, Swapnil Joshi, Solutions Architect at IBM India, illustrated how data pipelines must be designed for elasticity and agility. He detailed a typical AI-driven lifecycle: ingestion of raw data in various formats, curation via distributed storage systems, high-speed inferencing with in-memory computation and GPU acceleration, and the delivery of actionable insights.

Swapnil spotlighted a paradigm where storage systems not only manage data volumes but recognize micro-changes in customer behavior. This enables real-time updates to decision systems, ensuring customer interactions remain hyper-relevant across product lines and platforms.

FROM GATHERING TO ONBOARDING

Sanjay Agrawal, CTO of Hitachi Vantara, traced the evolution of enterprise data practices. What was once referred to simply as “data gathering” has matured into a highly engineered discipline he calls “data onboarding.” Unlike older systems that drew from a handful of sources, modern BFSI firms must ingest and process terabytes of data from myriad points, transaction logs, CRM systems, social media streams, IoT devices, APIs, and more.

Key to this process, according to Agrawal, is the adoption of metadata injection and automated extraction techniques. These allow institutions to compress what was once a months-long process into days, while ensuring consistency and integrity. However, the end goal is not just storage, it is business-readiness. For data to be considered onboarded, it must be sanitized, normalized, made discoverable through catalogs, and immediately usable by AI engines and decision systems.

USE CASES IN MOTION