TL;DR
A premium automotive manufacturer’s US post-sales organisation ran part replacements, warranty claims, service tracking, regional campaigns, reporting and customer notifications as separate, partly manual processes. Our engineer built Spring Boot microservices connected through Kafka events, with automated Excel/PDF reporting and containerised deployment on AWS, so a single service event drives tracking, notifications and reports without manual coordination.
Challenge
Post-sales work was spread across dealers and regions, and each step depended on someone noticing the previous one had happened. A warranty claim, a part replacement and the customer’s notification were handled in different places; regional campaign reporting was assembled by hand. The platform also had to meet production standards for a brand that cannot tolerate a failed customer notification or a lost claim.
Solution
Each workflow became its own Spring Boot microservice with clear ownership of its data. Services communicate through Kafka topics: when a service event is recorded, the tracking, notification and reporting services react to it independently, so a slow or failed consumer never blocks the others.
- One Spring Boot microservice per workflow, each owning its own data
- Kafka topics linking the services, so tracking, notifications and reports react to every service event independently
- Excel and PDF reports (Apache POI) generated from the same events
- Documents stored in S3 with controlled access
- Containerised with Docker, orchestrated on Kubernetes, deployed through Jenkins pipelines to AWS EC2
Stack and approach
Technologies
- Java 8
- Spring Boot
- JPA/Hibernate
- Apache Kafka
- Apache POI
- AWS S3
- Docker
- Kubernetes
- Jenkins CI/CD
- AWS EC2
Patterns
- One service per workflow
- Event sourcing of service activity
- Idempotent consumers
- Retry with dead-letter handling
- Query optimisation with JPA/Hibernate
- Production support rota with root-cause fixes rather than restarts
Results
Manual hand-offs between claims, parts, tracking and notifications were replaced by automated event flows. Regional campaign reports that were assembled by hand are generated from the platform. The platform ran in production for 14 months with the engineer on critical-issue support.
Delivered by CodaqLabs engineers while employed at previous companies; client names withheld under confidentiality.