Event-Driven ML System — Kafka × Image Classification
Production-style event-driven microservices for high-throughput ML inference
A horizontally scalable, event-driven architecture for async image classification at scale. A Node.js ingestion API publishes jobs to Kafka; a Python ML inference service (consumer group) classifies in parallel; a results service streams status back to a Next.js frontend. Fully Dockerized with NGINX, load-tested with autocannon, and tunable via Kafka partitions + replica scaling.