Loading…
Loading…
Scale journey
One city (~50K users)
Stage 1 of 5
A simplified teaching path inspired by public talks and posts — not an exact history of Uber. Stages are ordered the way you meet each problem, so some of what breaks is a teaching step rather than something Uber lived through.
Signature move
Uber's load isn't reads and writes — it's matching a rider to the nearest of thousands of drivers while everyone is moving. Driver locations stream in continuously and live in an in-memory geo-index, so a match is a spatial lookup over fresh data rather than a scan of a database. Everything downstream — surge pricing, dispatch, trip state — hangs off that stream of location updates.
Comfortable here. Drag the traffic up and watch what gives way first.
Now try building one of these architectures from a blank canvas — you're scored on the services you place and how you wire them.