Back to all selected work
Distributed Systems · Financial Engineering·2024·built
PMS — Trade Lifecycle Platform
Event-driven trade processing on Spring Boot, Kafka, and AWS EKS.
JavaSpring BootApache KafkaPostgreSQLAWS EKSDocker
Overview & Context
Processes high-throughput trade events across validation, matching, and settlement services with strict idempotency and auditability.
The Problem
Financial transaction pipelines cannot tolerate lost messages, dual-write inconsistencies, or unhandled consumer crashes during peak trading volatility.
The Architectural Approach
Engineered an asynchronous architecture using the Transactional Outbox pattern to eliminate distributed transactions. Deployed idempotent Kafka consumers with dead-letter topic failover, PostgreSQL persistence, and containerized workloads orchestrated on AWS EKS.
Key Technical Highlights
- Transactional Outbox pattern ensuring zero message loss between DB and Kafka
- Idempotent event consumers with dead-letter queue routing and retry policies
- Containerized microservices orchestrated on AWS EKS with Prometheus monitoring
Next Case Study
View projectCredit Card Fraud Detection ML
Anomaly detection research published in IEEE Xplore for ICCPCT 2025.