Statement Factory Data Platform

Turning raw banking transactions into compliant, ready-to-send statements

MADIS builds the cloud data platform that prepares the source data for a bank’s statement factory – on Azure and Databricks.

Summary

Regional market leader bank’s statement sends must be accurate, compliant and built from data spread across several core banking systems. MADIS designed and develops a cloud Data Lakehouse on Azure and Databricks that collects these transactions, standardizes them and delivers ready-to-use XML files to the bank’s Statement Factory.

The client

A Hungarian bank, which serves a large base of retail and corporate customers, all of whom receive regular bank statements. Behind each statement lies transaction data generated by many different banking subsystems.

The challenge

A bank statement is a regulated document: it has to be complete, accurate and formatted according to both legal obligations and the bank’s own specification. Before a single statement can be produced, transactions have to be collected from various core banking systems, categorised and brought into one standard structure.

The difficulty lies in the diversity of the sources and in the zero tolerance for error.

• Heterogeneous sources: event-driven and snapshot-based data arriving from multiple core banking systems.
• Strict compliance: every statement must meet legal requirements and the bank’s internal criteria.
• Several consumers: the prepared data must serve the Statement Factory and other systems connected through the bank’s middleware.

Our approach

We built a processing chain that takes transactions from their source systems all the way to print-ready statement data, in three main steps.

Step 1 – Ingestion (ETL)

Transaction data from the banking subsystems is loaded into Azure Data Lake and processed with Databricks and PySpark, where it is integrated and normalised regardless of whether it arrived as an event stream or a snapshot.

Step 2 – XML mapping

The standardised data is mapped into the structure required for statement generation and transformed into XML files in the camt.053 banking standard.

Step 3 – Export

The finished files are stored securely and made available to the Statement Factory, which produces the final statements, while other systems can access them through the bank’s middleware.

The solution

The result is a cloud-based Data Lakehouse that works as a data marketplace for statement generation, serving both retail and corporate customers. It turns raw transactional data into a single, standardised source that downstream systems can rely on.

The platform was designed with scalability and security in mind, so it can absorb new data sources and requirements as the bank’s needs grow.

Technologies: Azure Data Lake · Azure Databricks · PySpark · camt.053 XML

Results

• Statements customers can rely on: accurate, legally compliant statements that meet the bank’s own criteria.
• One standard data source: transactions from different subsystems are integrated and normalized in a single place.
• Ready for growth: a scalable architecture that supports further optimization and new use cases.

Facing a similar challenge?

Building a compliant, scalable data foundation for regulated reporting? Our data engineering team can help you design it.