Mini Ride-Share Platform (Spring Boot + Kafka + K8s)

This doc will be updated as I progress through implementation.

I'm going for an MVP implementation first, and then pivot to optimizing and refactoring. Initial focus is on getting my local infra setup (Helm/k8s/Github Actions) and barebone decoupled microservices and inter-service communication. Once that is done, I will move to resilience, fault tolerance and optimization aspects.

Project Summary

This project is a small ride-share style app I'm building to practice and demo that I can take something from idea to production-ish setup.

It covers the full flow:

• writing backend services in Java/Spring Boot,
• wiring async events with Kafka,
• storing core data in MySQL,
• write a location matching algorithm
• using Redis for fast lookups, and
• Finish with some Go components

Then I deploy everything on Kubernetes and manage releases the GitOps way with ArgoCD on a Bare-Metal cluster. Also, I will be adding Terraform for provisioning clusters in GCP and AWS to deploy this app.

Hardware - R720 PowerEdge Server x 4 (From my Homelab)

The goal isn’t to build a full Uber clone - it’s to show practical engineering skills across:

• service design and API development
• event-driven architecture
• caching + database usage
• containerization and K8s deployment
• CI/CD + GitOps workflows
• basic scaling patterns (replicas, Kafka partitions, stateless services)

You can think of it as a “real-world-ish” demo project that proves I can design, code, ship, and scale a distributed system with a modern stack.

---

Diagrams & Screenshots

Argo CD

Github Actions

---

Technical Details

TODO: Update README as I progress through implementation

App will (probably) contain

• auth-gateway-service
• rider-service
• driver-service
• dispatch-engine
• location-tracker
• common-dto

K8s related setup

• Enabled metallb for provisioning LBs
• Enabled microk8s dashboard
• Installed argocd
• brought in remote config from cluster host into laptop
• patched svc for k8s and argocd dashboards

  kubectl patch svc kubernetes-dashboard-kong-proxy -n kubernetes-dashboard -p '{"spec": {"type": "LoadBalancer"}}'

  kubectl patch svc argocd-server -n kubernetes-dashboard -p '{"spec": {"type": "LoadBalancer"}}'

Things to automate

• Configure mysql password, jwt secret, dockerhub as secrets outside of git
• Creating ingress-controller(nginx) & Patch it to type LB

kubectl apply -f https://raw.githubusercontent.com/kubernetes/ingress-nginx/controller-v1.10.1/deploy/static/provider/baremetal/deploy.yaml

kubectl patch svc ingress-nginx-controller -n ingress-nginx -p '{"spec": {"type": "LoadBalancer"}}'

Debugging

Notes on troubleshooting this setup: Debugging.

To Do

1. Move DTOs into a depedency model.
2. Move to a shared/overlayed Helm chart method since I have more services now.
3. Validating Admission Controllers
4. Create cursor skills for auto linting
5. Move Kafka to strimzi version.
6. Deploy an cache for images/dependencies etc.
7. Move the base images for deployment to minimal distroless images.
8. Add caching into Github actions for quicker build times.
9. Refactor to project level Helm chart from app level helm chart.
10. Tighten up CORS implementation.
11. Implement distributed locking for driver matching logic

Decisions :

1. Not using Spring security AuthenticationManagement for the MVP
2. Going with a shared secret on K8s for the jwt public key. Later move it into a JWKS endpoint on authservice and limit the public key to only one service. Need to think of trade offs there.
3. Not using an umbrella chart method and instead create a argo-cd app pointed at bitnami chart for Kafka.
4. Not using a image cache like Nexus/Artifactory.
5. Not using Transactional Outbox Pattern till I finish the basics