League of Basketball: league management software that ends the courtside score argument

A basketball game in Karachi runs on two tablets and a phone. One tablet holds the box score, the running tally of points, fouls and rebounds. The other holds the quarter clock. A fan in the stand follows the same match in an app. When those three disagree, the argument happens courtside.

League of Basketball is a Karachi organisation that runs men’s, women’s and varsity competition in Pakistan. A system of record is the one place a fact is allowed to be true, and in August 2023 the league had none. Both server repositories open on blank framework templates. Across 257 commits there is no import script and no adapter for an older system.

Every figure here is read from the git history of three repositories: the backend, the operator console and the fan app. The first commit lands on 1 August 2023, the last on 7 January 2026.

If you run an operation where two people touch one record while it is changing, you already know this failure. The screens drift apart. Whoever speaks loudest becomes the source of truth.

Three things were true courtside. Scorers mis-enter points and have to take them back. A finished box score should stop being editable. Venue Wi-Fi drops mid quarter.

It is the same problem as an order living in a warehouse tool, a spreadsheet and a customer email at once. The fix is deciding which system is allowed to be right.

League management software is the software around a live game: fixtures, teams, players, the score and the result. Autonomous Technologies builds and then runs custom operational systems for the person accountable after launch. Here that meant all three surfaces.

The backend holds leagues, seasons, teams, squads, players and matches. On top of it runs a real-time layer built with Django Channels and Redis, tools for keeping live connections open. One connection stays open per match and pushes every update to every screen watching it. In plain terms, each tablet and each phone reads one clock and one score.

The clock is not a timer ticking in an operator’s browser. The server stores a start time and the elapsed time already banked, then works out what is left. It refuses to start a quarter unless five players a side are marked on court, and it names the team that is short. When a quarter reaches zero it stops itself and moves on.

Two engineers carried this for its whole life. Usama wrote the backend and the scoring engine, 79 of its 96 commits. Saif wrote the fan app in React Native, one codebase for both phone platforms, 110 of 116. Abdullah handled infrastructure and review. The console was built for a real game, not a tidy one. A correction mode pulls a wrongly entered point back out. The match goes read-only at full time. Nobody builds a reconnect button for a network that behaves.

Then the league went quiet. It paused the project for its own reasons. The backend went nine and a half months without a single commit.

In September 2025 it came back with a request that touched everything. Google and Apple both require an in-app route for a user to delete their account. Meeting that meant a delete path in the API, a new database model, an admin screen, a support email, a settings screen and a release build. Cross-stack, in other words.

That is the real question about an outside engineering team. Not whether they can build it. Whether, a year later, anyone still understands your system. The same two engineers took this from first commit to merged release in 24 days, between 10 September and 4 October 2025. No discovery phase. No rewrite. Eleven months on, none of those files has needed a fix.

The app is live on Google Play, not on the App Store. The engagement was paid at a discount, because the league is a non-profit. The repositories are the client’s own.

This fits you if two people arguing about one record costs your team hours of rework, and the record keeps changing while they argue. Courtside scoring, dispatch boards, live stock counts, anything with a clock on it. It is the work behind custom engineering and integrations, and the point of it is fewer hours lost to rework.

Walk away if you need a certified, test-gated, audit-ready process from day one. This build did not have one. The scoring engine has no automated tests. Two of the three front ends have no continuous integration, so no automatic check runs before a change ships.

What we would do differently is write the system down: no runbook and no architecture document exist. Those 24 days worked because Usama and Saif were both still here. Had either left, the client would have paid for a rediscovery project. A written system map and tested critical paths are now scope, not a nice-to-have. That is how we ran the embedded product team for Memox and the long retainer behind The Republic’s bilingual publishing platform.