CareSRE

Outpatient departments in government hospitals run on physical queues. A patient arrives early, waits without knowing where they are in the line, and has no way to find out short of asking someone. The staff have the same problem in reverse: no reliable view of load until the corridor is already full.

The judging criteria for the hackathon were public, and they were not about architecture. They asked whether the problem was real, whether a judge could follow the solution quickly, and whether a hospital could plausibly adopt it. Roughly 2,400 teams entered. Most of them read those criteria and then built the most technically ambitious thing they could finish in the time.

We built the smallest system that made the queue legible to both sides: intake, position, and expected wait, with the AI doing triage prioritisation rather than being the demo. Then we spent the time we had saved on the part that was actually being scored: the user journey for a patient and for an administrator, and the case for why a hospital would say yes.

The first version tried to predict wait times precisely. It was a good modelling problem and the wrong product decision. A patient does not need a number accurate to the minute, they need to stop being in the dark. Once we replaced the prediction with an honest range, the feature got simpler and more useful at the same time.

1st place from a field of roughly 2,400 teams, on the strength of the journey and the adoption case rather than the architecture.

The game everyone is visibly playing and the game being scored are usually two different games. Technical ambition is the easiest way to feel productive while working on the wrong one.