An End-to-End Data Quality Management Framework for Banking Systems with Generative AI Integration

Author: Debabrata Pruseth
Publication Date: 2025/06/09
Document Type: Technical Note / Research Article
Language: English

Abstract

Data quality management is a foundational capability for banking systems because financial institutions rely on trusted data for customer onboarding, transaction processing, regulatory reporting, risk aggregation, anti-money laundering monitoring, credit decisioning, operational resilience, and enterprise decision-making. Poor-quality data can weaken customer trust, increase regulatory exposure, distort risk reporting, and reduce the reliability of downstream analytics and artificial intelligence systems. In modern banking environments, data is created, stored, transformed, used, archived, and deleted across a complex network of core banking systems, customer platforms, transaction engines, risk systems, compliance tools, finance systems, data warehouses, data lakes, and end-user computing assets. This complexity creates persistent challenges around accuracy, completeness, consistency, timeliness, validity, lineage, ownership, and traceability.
The proposed research-style framework extends that lifecycle model into a formal banking architecture and governance approach. It positions Generative AI as an augmentation layer, not as an autonomous data controller. Generative AI can support data stewards, business owners, technology teams, risk managers, compliance teams, and operations users by summarizing issues, interpreting policies, generating candidate rules, explaining lineage, and drafting evidence-based narratives. However, because banking data is sensitive and regulated, GenAI integration must be governed through human review, retrieval-augmented generation, role-based access controls, audit logging, data masking, model risk management, and clear accountability.

The paper concludes that banks should treat data quality management as both a regulatory capability and an AI-readiness capability. Trusted data is not only required for accurate reporting and compliance; it is also essential for safe and scalable Generative AI adoption.

This paper presents an end-to-end Data Quality Management Framework for banking systems with Generative AI integration. The framework builds on the author’s original blog article, which organizes data quality management across six lifecycle phases: data creation, storage, processing, usage, archival, and deletion. The blog proposes practical Generative AI applications across these stages, including intelligent data entry assistance, metadata generation, automated data classification, lineage documentation, anomaly interpretation, report drafting, policy analysis, privacy scanning, retention planning, archive summarization, erasure request orchestration, and deletion audit summarization.

Keywords
data quality management, banking systems, Generative AI, GenAI, data governance, data lifecycle, data creation, data storage, data processing, data usage, data archival, data deletion, BCBS 239, GDPR, metadata, lineage, data classification, data stewardship, AI governance, retrieval-augmented generation, regulatory reporting, operational risk

Download Research PDF

An End-to-End Data Quality Management Framework for Banking Systems with Generative AI Integration

Suggested Citation
Pruseth, D. (2025). An End-to-End Data Quality Management Framework for Banking Systems with Generative AI Integration.

Companion Note
This page provides the abstract and full-text PDF for the research version of the article. A companion blog post explains the same work in a more narrative and implementation-focused style.

Read the companion blog:
https://debabratapruseth.com/end-to-end-data-quality-management-framework-dqmf-in-banking-with-genai-integration/