Event-driven microservices processing millions of points-earning and reward-redemption events daily with zero data inconsistency. Kafka-based architecture decouples domain services, enabling 99.99% uptime with multi-region resilience and replay-safe failure recovery.
🎯 Overview
Klarna’s loyalty platform handles fan-out of purchase events to points crediting, tier recalculation, partner notification, and analytics. Decoupled Kafka consumers prevent service failures from cascading, while consumer group offsets guarantee exactly-once delivery even during restarts.
🛠️ Tech Stack
- Kafka: Durable event bus with consumer groups ensuring exactly-once delivery and replay capability.
- TypeScript: Type-safe event schema definitions preventing silent schema-mismatch bugs.
- Python: Analytics pipeline processing raw event streams into engagement metrics.
- AWS: Multi-region deployment (MSK, RDS Aurora global clusters) enabling 99.99% uptime.
📈 Key Features
- Event-driven architecture with Kafka decoupling independent domain services
- Exactly-once delivery semantics preventing double-credit incidents
- Real-time engagement metrics and loyalty tier calculations
- Multi-region deployment with automatic failover
- Kafka replay enabling caught-up recovery without data loss