Projects / Product · 2026
ApiPulse
API monitoring and analytics for Node.js services.
- SDK
- Ingest API
- RabbitMQ
- Worker
- OpenSearch
- Dashboard
01 — The problem
When an API slows down or starts failing, the first questions are which route, since when, and what the failing requests looked like. Answering them usually means a heavy monitoring suite or digging through logs.
02 — The solution
ApiPulse is a small SDK plus a processing pipeline. The SDK sits inside an Express or Fastify app, measures every request and sends batches to an ingest service. A queue and a worker do the storage work, so the monitored app never waits on a database.
- TypeScript
- Node.js
- Fastify
- RabbitMQ
- OpenSearch
- MongoDB
- React
- Docker
The repository is private. I'm happy to walk through it on a call.
03 — What it does
- Tracks total requests, error rate and average latency per route.
- Filters by time range (last 15 minutes to 7 days) and by HTTP method.
- Shows traffic and latency on one chart so a spike can be read against load.
- Keeps the stack trace with the request that caused the error.
- Bring your own storage: connect your own OpenSearch cluster, with credentials encrypted using AES-256-GCM.
- Rate limits and payload size limits protect the ingest endpoint.
04 — How it's built
- SDK
- TypeScript middleware for Express and Fastify. Batches events and flushes every 2 seconds, with sampling and data masking.
- Ingest
- Validates the project key and payload, then pushes the batch straight to RabbitMQ.
- Worker
- Consumes from RabbitMQ and bulk-indexes into OpenSearch. Caches storage clients in memory for 10 minutes.
- API
- Authentication, projects and API keys in MongoDB; queries OpenSearch to serve dashboard metrics.
- Dashboard
- React and Vite app with Recharts and Tailwind CSS, light and dark.
Next project
Counselling Platform →