Seylan Bank Customer Segmentation & Rewards Platform

The Challenge

Seylan Bank required a sophisticated customer segmentation and rewards management system to handle millions of customer transactions and events daily. The platform needed to:

• Automatically categorize customers into different profiles (Retail, Corporate, Elite, Prime, etc.) based on complex business rules
• Process real-time events from multiple banking systems (Core Banking, ATM transactions, account activities)
• Calculate benefits and charges dynamically based on customer profiles and transaction history
• Provide 360-degree customer views for relationship managers and customer service teams
• Enable business users to modify segmentation rules without requiring code deployments
• Handle data extraction from legacy DB2 systems and modern MySQL databases
• Support quarterly and monthly segmentation processes with recovery mechanisms
• Generate reports and CSV files for core banking integration

The existing manual processes were time-consuming, error-prone, and couldn't scale to handle the bank's growing customer base and transaction volume.

The Solution

Architecture Design

Architectural Decisions

1. Modular Microservices Architecture: Separated concerns into independent modules (Event Submission, Segmentation Engine, Data Loader, Server API) allowing independent scaling and deployment.

2. Event-Driven Processing: Implemented asynchronous event processing pipeline enabling real-time customer profile updates without blocking the main transaction flow.

3. Rule-Based Engine: Integrated Drools rule engine allowing business users to modify segmentation rules through UI without code changes, reducing deployment cycles from weeks to hours.

4. Multi-Database Support: Designed abstraction layer supporting both legacy DB2 (AS400 iSeries) and modern MySQL databases, enabling gradual migration while maintaining backward compatibility.

5. GraphQL API Design: Chose GraphQL over REST for flexible data fetching, reducing over-fetching and enabling frontend teams to request exactly the data needed for each view.

Key Contributions & Problem Solutions

Tech Stack

Impact & Results

Business Impact

• Operational Agility: Reduced segmentation processing from hours to minutes, enabling faster time-to-market for campaigns.
• Customer Experience: Real-time 360-degree views empowered agents to solve customer issues instantly.
• Cost Reduction: 90% reduction in manual processing costs through automation of stored procedures and reporting.

Technical Efficiency

• Performance: 70% reduction in query execution time ensured system stability under high load.
• Scalability: Modular microservices architecture allowed independent scaling of ingestion and reporting layers.
• Modernization: Migration to React/TypeScript and GraphQL significantly improved developer velocity and code maintainability.

Operational Excellence

• Reliability: Automatic recovery and retry mechanisms reduced manual support incidents by 90%.
• Data Integrity: Robust ID management and migration tools guaranteed zero data loss for millions of records.

Key Learnings

1. Event-Driven Architecture: Learned the importance of asynchronous processing for high-volume systems. Implementing event-driven patterns enabled the system to handle millions of events without blocking.

2. Database Optimization: Gained deep expertise in query optimization, indexing strategies, and batch processing. Understanding query execution plans and database internals was crucial for performance improvements.

3. Rule Engine Integration: Learned to design systems that allow business users to modify behavior without code changes. This required careful abstraction and configuration management.

4. Migration Strategies: Developed expertise in data migration, especially for customer ID systems. Learned the importance of history tracking, rollback capabilities, and gradual migration approaches.

5. Full-Stack Development: Gained experience in both backend (Java/Spring) and frontend (React/TypeScript) development, understanding how to design APIs that serve frontend needs efficiently.

6. Testing & Quality: Learned that high test coverage is not just about numbers but about having meaningful tests that catch real issues. Integration testing with Spring was particularly valuable.