Home/Engineering blog
Engineering
How we build Yusr
There is no branch to fall back on. If the app is down, the bank is closed. That constraint shapes almost every decision we make, and this is where we write it down: the architecture, the numbers, and the things we got wrong the first time.
What you'll find here
Posts written by the people who carry the pager, not by our marketing team. Expect real latency figures, real incident timelines, and the occasional admission that version one was a mistake.
Something here look wrong? Tell us at engineering@yusr.com.sa. For anything about your own account, the help centre is much faster.
64
Engineers, designers and SREs
2
Cloud regions behind the app
Weekly
Release train, every Tuesday
Q1 2025
First production deploy
Latest posts
Running a bank across two regions
Yusr runs from a primary region in Riyadh, with a second region in Frankfurt kept warm and continuously replicated. How the data gets there, what a failover actually looks like from the inside, and why we rehearse it on a Tuesday morning instead of waiting for a bad night.
Omar Al-Zahrani · Chief Technology Officer · 9 min read
How we score risk in real time
Every card authorisation and every transfer is scored before it completes, inside a budget of forty milliseconds. A walk through the feature pipeline, the score bands we act on, and what happens to a payment that lands in the top band.
Hessa Al-Otaibi · Staff Engineer, Risk Platform · 11 min read
Instrumenting the Yusr app without slowing it down
We wanted to know which screens people abandon and where the app feels slow, without adding a second to cold start. Batching, sampling and a lot of profiling later, the analytics layer costs us under twelve milliseconds on launch.
Tariq Al-Ghamdi · Senior Engineer, Mobile · 8 min read
From the first year
Posts from before we moved the blog onto the main site. They're being reformatted one at a time and will link up here as they land.
| Published | Post | Author |
|---|---|---|
| 24 July 2025 | Moving the ledger without stopping the bank | Saud Al-Rasheed |
| 6 May 2025 | What eight weeks of mada integration testing taught us | Lina Al-Harbi |
| 13 March 2025 | On-call at a bank that never closes | Maha Al-Juhani |
How we work
Four habits, held to fairly tightly
None of this is novel. It is the unglamorous set of practices that keeps half a million people able to pay for groceries on a Thursday evening.
See the live system statusSix teams, each owning a slice end to end: onboarding, ledger, cards, transfers, risk, and the app itself. Whoever builds a service carries the pager for it. The rota is shared, paid, and capped at one week in six so nobody burns out quietly.
A resilience plan nobody has executed is a document, not a plan. We run a full regional failover exercise every quarter, in working hours, with the whole team watching. The first one took us well over an hour. The most recent took eleven minutes.
We buy identity document checking, message delivery and product analytics from specialists who do nothing else. We build the ledger, the risk engine and the app, because those are the things customers actually experience as Yusr.
Every vendor gets the same review: what it costs, what it breaks if it disappears overnight, and how quickly we could replace it.
Nothing goes out without a number attached: cold start time, authorisation latency, the share of applications that reach the end. If a release moves one of those the wrong way, it comes back out the same day. Rolling forward is a decision, not a default.
Who writes here
Omar Al-Zahrani
Chief Technology Officer
Joined before the licence came through and hired the first eight engineers. Spent fifteen years in payments infrastructure across the Gulf beforehand. Writes mostly about architecture and the cost of getting resilience wrong.
Hessa Al-Otaibi
Staff Engineer, Risk Platform
Owns the scoring service that sits in front of every payment. Came from a background in streaming systems and has strong opinions about latency budgets, feature freshness, and the difference between a model and a decision.
Tariq Al-Ghamdi
Senior Engineer, Mobile
Works on the iOS and Android apps, mostly on start-up time and the onboarding flow. Keeps a public leaderboard of every millisecond removed from cold start since launch, and is not shy about defending it in review.
We're hiring
If you read this far, you'd probably fit in
We're adding engineers across the ledger, risk and mobile teams, and a couple of roles that don't fit neatly in any of them. Small teams, real ownership, and a product people use several times a day.
What our interview looks like
A conversation. Forty-five minutes on what you've built and what you'd want to build next.
A real problem. A scaled-down version of something we actually hit, worked through together. No whiteboard trivia.
The team. Time with the people you'd sit with, including whoever shares your on-call week.
Most candidates hear back within five working days. Roles are based in Riyadh unless the listing says otherwise.