Back to case studiesBackend / Distributed Systems
PulseQueue — Live Distributed Job Queue Platform
A real durable job queue you can watch work — submit a job and see it flow queued, processing, retried with exponential backoff, or permanently dead-lettered, live.
Next.jsTypeScriptUpstash QStashUpstash RedisVercel
Real retries, real DLQ
View Live DemoThe Problem
Most portfolio demos fake distributed-systems behavior with client-side timers. That doesn't prove anything about how someone reasons about queues, retries, backoff, or failure recovery — the things that actually break in production backend systems.
Our Approach
- 1Architected the queue on Upstash QStash — a genuinely durable HTTP message queue — instead of simulating retry logic client-side, so backoff and delivery guarantees are real, not scripted.
- 2A 'Chaos Mode' toggle makes the worker endpoint intentionally fail on a random subset of jobs, letting QStash's own retry engine react exactly as it would to a real outage.
- 3Hardened the worker callback with QStash signature verification and idempotent job-state updates, so a forged or duplicated delivery can't corrupt job state.
What This Achieves
- Visitors watch a real message queue retry, back off, and dead-letter a job in real time — not a mocked animation.
- A dead-letter queue panel reads directly from QStash's own DLQ API, with one-click replay.
- The architecture (durable queue + stateless worker + idempotent callback) is the same shape used by production job systems at scale.
Next Custom Engineering Case Study
Immutable Blockchain Certificate Verification System
Have something similar in mind?
Scope Your Project