Work · CS-001
Where the whole city goes out
A live events platform in Central America: artists, venues, fans and tickets in one place
Problem
Two products that want opposite things, growing at once. A social network for going out: artists announce, fans follow, review and recommend, and the feed has to be alive every second and forgive anything. And ticketing: one seat sold exactly once, and a gate that has to say yes or no in under a second with no signal. Most platforms pick one and bolt on the other. The client had both, and every big launch was a fight between them: an artist's announcement would flood the feed and take the on-sale down with it.
Solution
We separated the physics and joined the data. Feed, reviews and recommendations run on an architecture that scales sideways and tolerates being a few seconds behind. Inventory and gates run on one that is strictly consistent and keeps working with the network cut. Between the two, one graph: artist, venue, event, fan, ticket. A scanned ticket becomes proof of attendance, so a review comes from someone who was in the room, and the next recommendation comes from where people actually went, not from what they said they liked. The artist sees an audience of people, not a count of tickets.
Outcome
On-sales stopped taking the feed down, and the feed stopped taking on-sales down. Reviews carry the weight of a scanned ticket. Artists announce to the fans who came last time. The platform runs the city's biggest nights and its quietest bar on the same graph.
What we left behind
One graph of who plays, who goes and what they said, and two architectures that never had to become the same.
If your operation looks like this, we should talk.
Tell us what yours looks like. We reply within 24 hours.