Built using Clean Architecture and CQRS via the Mediator library.
Implemented real-time notifications using SignalR with an in-memory message queue for asynchronous processing.
Implemented end-to-end observability using OpenTelemetry, collecting distributed traces and metrics, exposing them to Prometheus, and visualizing request throughput, latency, and error rates through Grafana dashboards.
Used Redis caching with event-driven cache invalidation to keep data fresh and responsive.
Integrated Serilog logging with Elasticsearch and Kibana for log analytics and diagnostics.
Used cursor-based pagination for efficient large dataset retrieval over offset pagination.
Deployed to Azure using GitHub Actions CI/CD pipelines.
Implemented Token Bucket and Sliding Window Counter algorithms via atomic Lua scripts.
Designed dual-mode architecture: transparent reverse proxy and standalone check API, allowing drop-in integration with any stack.
Implemented per-route configuration with fail-open/closed resilience policy for Redis outages.
Instrumented full observability pipeline using OpenTelemetry for distributed tracing, Prometheus for metrics, and Grafana for dashboards.
Shipped production Kubernetes manifests using Kustomize to deploy the full stack (Limitry, Redis, OTel Collector, Jaeger, Prometheus, Grafana) into a dedicated namespace with a single command.
Load-tested with k6 across smoke, load, and spike scenarios (180K+ total requests, p95 latency under 6ms for Check API and 35ms for Proxy mode, 0% error rate at 100 concurrent users).