Designing a Self-Service Data Platform: How Data Mesh is Modernising Finance

The Challenge: The Bottleneck of Centralised Data Architectures

Many organisations, particularly in the heavily regulated financial sector, struggle with the same problems:

• Data silos: Valuable information is trapped in isolated systems.
• Long waiting times: Business units often wait weeks for reports from central IT teams.
• Lack of transparency: There is no company-wide data catalogue and no clear responsibilities (data ownership).

Traditional, centralised data architectures reach their limits here. Our goal was therefore to design a platform that resolves these bottlenecks and establishes the mindset of “Data as a Product”.

The Solution: A Data Mesh Approach

Instead of pumping all data into a central lake, the data mesh approach shifts responsibility back to the business domains. The idea: those who generate the data know best how to use and maintain it.

For the financial institution, we developed an architecture based on the following pillars:

• Decentralised Data Ownership: Business units offer their data as products.
• Self-Service Infrastructure: A platform that enables teams to provide and consume data without deep technical expertise.
• Federated Governance: Global standards ensure security and interoperability without slowing down agility.

The Proof of Concept: The «360-Degree Customer View»

Theory is good, practice is better. To prove the feasibility of our concept, we defined a concrete use case – the 360-degree customer view – as a Minimum Viable Product (MVP). The goal was to aggregate customer data from various sources securely and compliantly to enable holistic customer service.

Technology Deep-Dive: Open Source as a Foundation

Our evaluation and cost-benefit analysis identified a combination of powerful open-source components as the best solution for true self-service capability:

• Apache Iceberg: As an open table format, it enables high-performance analytics on vast data sets and prevents vendor lock-in.
• Trino: This federated query engine allows SQL queries across different data sources without the need to copy data.
• Apache Kafka & Flink: For real-time data integration and processing.

This architecture not only offers a high degree of automation but also lays the foundation for future scalability.

Conclusion: Agility Through Decentralisation

The results of our work clearly show: simply optimising existing tools is often not enough. Switching to a data mesh concept with a dedicated governance engine enables organisations to deliver data products faster and dramatically reduce the “time-to-market” for analytics.

At Swissware IT, we understand that technology must always serve a purpose: to simplify processes and create real value. This project underscores our expertise in analysing complex technical challenges and designing future-proof solutions – whether at scale for financial institutions or tailored for your SME.

Project authors: Giuseppe Scavetta, Marco Stomaci, Mirjam Islamovic & Nevzat Abduli (2025)