Challenge
The system was spending too much time in a CPU-bound path, which made every request feel slow under load.
Solution
I profiled the hot path, moved expensive work off the event loop, and used Worker Threads to keep the API responsive.
What I Built
- Profiling to isolate the slowest parts of the request flow
- A CPU-heavy workload moved off the main event loop
- Worker Threads to keep the API responsive under pressure
- Load testing to compare before and after performance
- Documentation of the trade-offs behind the optimization
Impact
The final result was a 98% latency reduction and a much clearer picture of how to diagnose Node.js performance issues.
Why It Matters
This project shows that I can turn a vague slowdown into a measurable engineering improvement.