OnRide

A ride-hailing backend built with Spring Boot microservices, Kafka, gRPC, and H3 geospatial matching, running on AWS EKS.

Live demo

Swagger: https://unable-absolutely-packs-blogging.trycloudflare.com/webjars/swagger-ui/index.html

Jaeger: https://unable-absolutely-packs-blogging.trycloudflare.com/jaeger

Deploys are automated via GitHub Actions on every push to main.

Design

• Matching — ride requests and available drivers are batched on a fixed interval and assigned using the Hungarian algorithm.

• Geospatial lookups — H3 hexagonal indexing for nearby-driver queries.

• Inter-service communication — gRPC for synchronous calls, Kafka + Avro + Schema Registry for async events.

• Tracing — requests are traced across HTTP, gRPC, and Kafka via OpenTelemetry, viewable in Jaeger.

• Deployment — Kubernetes on AWS EKS, nginx Ingress, native k8s service discovery (no Eureka in the cluster).

  

Tech stack

Java 26 · Spring Boot 4 · Spring Cloud Gateway · PostgreSQL + Flyway · Redis · Kafka + Avro · gRPC · Uber H3 · OpenTelemetry + Jaeger · JWT · Docker (Jib) · Kubernetes (EKS) + nginx Ingress · GitHub Actions

Architecture & user flow

api-gateway → auth-service, location-service, ride-service, matching-service. discovery-service is local-dev only (no Eureka in k8s).

Rider: signup/login → POST /rides/quotes → POST /rides/book → GET /rides/matches. Driver: signup/login → /drivers/me → POST /locations/ping → GET /rides/matches → POST /rides/{id}/accept.

All under /api/v1/... via the gateway.

Running it locally

Prereqs: Java 26, Postgres, Redis.

1. docker-compose up -d — Kafka, Schema Registry, Kafka UI, Jaeger
2. Start Redis
3. Copy each service's .env.example → .env, fill in
4. Start discovery-service, then ./gradlew bootRun per service

Gateway: localhost:8080 · Swagger: /webjars/swagger-ui/index.html · Kafka UI: :8090 · Jaeger: :16686

Future scope

• Real-time surge pricing
• Metrics dashboards (Prometheus + Grafana)
• WebSocket layer for live location/match updates
• Spring Cloud Config