Make a phone call
work from anywhere.

What the work is actually like

Much of it is unglamorous and quite deep. You will chase a hundred and fifty milliseconds of added latency through three networks you do not control. You will work out why one carrier in one country returns a busy signal for a number that is ringing perfectly well. You will build the billing path that has to agree, to the cent, with a figure a user watched tick upward on their own screen — because on this product the receipt is the promise, and a rounding error is a broken one.

In exchange, the work is unusually legible. Either the call connected or it did not. Either the number on the receipt matches the number on the meter or somebody is about to write to support. There is very little room here to be busy without being useful.

How we work

• Small team, wide scope. You will own a region, a carrier relationship or a subsystem end to end — how it routes, what it costs, how it fails, and the tickets it generates when it does.
• Written over synchronous. Decisions get written down. It is the only way a distributed team stays coherent about something as detail-bound as voice routing.
• Money bugs are the highest severity we have. An overcharge of two cents is not a two-cent problem. It is the only claim this product makes, broken.
• If it needs an explanation, it is not finished. Our users are people under stress in a country that is not theirs, sometimes on a borrowed computer. Anything that only works once someone explains it counts as broken.
• Everyone talks to customers. The person who built the routing answers the ticket about the routing. It is the shortest path from a confused user to a fixed product.
• Having needed this yourself is a qualification. If you have lived abroad and spent an evening trying to reach an office back home, you understand the product in a way no amount of research transfers.