Real-time Systems

We build systems that process events as they happen rather than in nightly batches: live position tracking, order status changes, price updates, sensor data streamed to a dashboard within seconds. Our maritime anomaly-detection engine, an in-house research project, tracks live AIS vessel positions and flags close-approach patterns using geohash-based proximity detection, built on Kafka and Neo4j.

We design the event pipeline around what has to happen in seconds versus what can wait: WebSocket connections push updates to a browser in real time, while Kafka queues and buffers events between services so a slow consumer does not block the rest of the system. Where relationships between entities matter, such as which vessels are connected, we use a graph database instead of forcing that structure into relational tables.

• Kafka-based event streaming between services
• WebSocket connections for live dashboard updates
• Graph-based relationship and proximity detection
• Monitoring and alerting on event pipelines

Technology: Java 21, Spring WebFlux, Kafka, Neo4j, WebSocket, AWS.