Coding problems and system design boards that are ready the second you open them. Work alone, or with a guide watching and a voice on the line.
Invite-only while we’re small. We add people a few dozen at a time.
Turn on the live guide and it follows along as you code, the way a good interviewer sitting next to you would. When you drift, it says so, out loud and briefly, then lets you get on with it. Answer it, ask for a hint, or keep typing.
Pick a track or a timed set. Run checks the samples while you think; Submit runs the hidden suite and times your code on fixed hardware, startup excluded.
Runtime you can compare. Every submission is timed on the same dedicated cores, several runs, median kept, compile and startup left out.
Growth, measured. We run your solution on growing inputs and tell you how it scales, which is what the interviewer will ask.
Timed sets. Two problems, 45 minutes, no hints unless you ask. The closest thing to the real thing without a stranger on the call.
Sketch the architecture on the board, choose a load profile, and watch where requests queue. You find the bottleneck before the interviewer asks about it.
Results are instant and the same every time, so you can change one thing and see what it did.
Start a voice call with an interviewer that asks follow-ups, pushes on the weak parts of your answer, and gives you room to think. Switch to text whenever you like.
Calls are optional. What we keep afterwards is a short set of notes for your feedback, not the recording.
Every session is scored on the same six things, with a line of evidence for each score. After a few weeks you know where you are strong and what to practice next, and so do we.
Last session: missed the empty-input case and didn’t check for duplicate keys.
Practicing means getting things wrong, often. You should be able to do that without it following you around.