Backend systems
APIs, jobs, data models and services designed for reliability under load.
- Engagement
- Fixed-scope build, or an engineer embedded in your team
- Typical timeline
- 12 weeks
- Starts with
- A walkthrough of the one operation you are most afraid of running twice.
The system behaves perfectly until two people touch the same record in the same second. Then a seat is sold twice, a balance goes negative, and the logs show nothing at all, because nothing crashed.
Infrastructure that stops being the reason things break.
What we build
Concrete artefacts, handed over and documented.
- 01
A data model where the rules the business depends on are expressed as constraints rather than as hope
- 02
Transaction and locking design for the handful of operations where two users genuinely collide
- 03
Background jobs that are safe to run twice, with a queue, retries and a backlog anyone can see
- 04
Traces through every request, grouped errors, and alerts on queue depth rather than on server count
- 05
A load and failure suite that reproduces month-end volume before month-end arrives
What changes
Double bookings, duplicate charges and negative balances become impossible rather than merely rarer
A repeated job costs nothing, so recovering from an outage is a replay instead of an investigation
More traffic stops meaning more incidents
How it runs
- Weeks 1–2
01Model and failure cases
We list every operation two people can perform at once and write down what correct means for each of them.
- Weeks 3–9
02Build
Data layer, queues and jobs, with the rules held in the database rather than in whichever code path happens to run first.
- Weeks 10–12
03Break it deliberately
Concurrency tests, a flooded queue, and a dependency pulled offline mid-transaction, before a customer does it for us.
Chosen per project. Named here so you can see the shape of it.
- TypeScript
- Node.js
- Postgres
- Redis
- BullMQ
- OpenTelemetry
Questions we get asked
Tell us the outcome, not the tooling.
Send us the situation you are in. We will tell you which discipline it belongs to, what we would do first and what it costs, including when the answer is to wait.