Healthcare platform.
Distributed microservices + iOS, in production.
problem
A regulated healthcare platform needed to coordinate dozens of independent services, expose them safely to a native iOS app, and add AI features without rebuilding the backend each time.
approach
Contributed across the distributed system: backend services in multiple languages, all speaking gRPC for low-latency request/response and Kafka for event-driven workflows. Everything lives in Kubernetes, with a Swift iOS client at the edge. AI features were added as new services rather than bolted onto the monolith.
outcome
Shipped to production. The system handles 14k events/sec on Kafka with sub-40ms p99 latency on gRPC. New AI features can be added as isolated services without touching the core. (Specifics under NDA.)
stack rationale
- gRPC over RESTStrict schemas catch breaking changes at compile-time across multiple language services. Streaming RPCs simplify long-lived connections.
- Kafka for eventsDecouples services. Replaying events lets downstream consumers be added later without coordinating with producers.
- KubernetesMany teams, many languages, one deployment surface. Resource isolation per service, rolling deploys, no snowflakes.
full stack
- Java
- Python
- Go
- Swift
- gRPC
- Kafka
- Kubernetes