Timed coding interview
Three problems, a window sized to each one, and the constraints of a real screen: a limited number of runs and submissions, a penalty for every failed submission, and hints that cost you marks. A report at the end shows exactly where they went.
Choose a set
The rules
- Easy problems get 12 minutes, medium 20, hard 30.
- 8 runs and 3 submissions per problem.
- A failed submission costs 15 marks; opening a hint costs 10.
- Opening the solution scores the problem zero, as it would in a real screen.
Write a solution and run it against the real test table.
Why a timer changes the problem
Most people practise by solving a problem and stopping when it works. A real screen is a different task. You are being watched while you decide what to try, you cannot run the code as often as you like, and a wrong submission is not free. The skills those constraints test are not the ones that untimed practice builds.
The commonest failure is not running out of time. It is spending eleven minutes on an approach that was never going to work, because there was no moment at which the candidate stopped and asked whether it would. In a timed run you feel that moment arrive. That is the point of it.
How this one is scored
The window is set by difficulty — 12 minutes for an easy problem, 20 for a medium, 30 for a hard one — which is roughly what a screen allows per question once the conversation is stripped out.
- Runs are limited. 8 per problem. Interviewers notice candidates who compile-and-pray instead of reading their own code, and the limit makes you read it.
- Submissions cost. 3 per problem, and a failed one costs 15 marks. Submitting to find out whether it works is a habit worth breaking before someone is watching.
- Hints cost 10 marks. Taking one is usually still correct — a hint you ask for early beats fifteen silent minutes — but it is not free, exactly as in the room.
- Opening the solution scores zero for that problem, which is also what happens when you are told the answer in a real interview.
The report at the end separates the two things that matter: whether you got there, and what it cost you to get there. A pass with three failed submissions and two hints is a different signal from a clean one, and the report says so rather than averaging them away.
How to use it
Work the course first. Sitting a timed set before you know the patterns measures something real but teaches you very little, and the frustration is not instructive.
Once a pattern is familiar, take that pattern's set under the clock. When the mixed screen stops producing surprises you are close to ready. Most people need several attempts, and repeating a set you have seen is still useful — the second time you are practising the decision, not the recall.