Ask for code without standing up a compiler.
Most hiring questions do not need a program that builds. They need to know whether somebody can write the query. Ask for a few lines, get them marked on the approach, and skip the runner entirely.
SELECT * FROM (
SELECT o.*, ROW_NUMBER() OVER (
PARTITION BY customer_id
ORDER BY created_at DESC) rn
FROM orders o
) t WHERE rn <= 5;Marked on approach, with the reasoning on the report.
The coding question that fits inside a test.
No compiler, no test runner
Nothing to install on your side or theirs, no language runtimes to pick, no sandbox to wait on. The question is written in the same editor as every other question on the paper.
Short answers, not take-homes
A loop, a query, a function signature, the fix for a broken snippet. The size of answer somebody can give inside a timed test, in any of 19 languages.
Marked on the thinking
The AI reads what they wrote against what the question asked for and scores the approach, with the reasoning on the report. Any score is yours to override.
Short coding questions, answered.
No. That is the point of it: there is no compiler and no test runner in the loop, so there is nothing to configure and nothing to wait for. The answer is read and marked against what the question asks for.
The same way a written answer is: AI grading marks it against the question and the marks you set, and shows the reasoning next to the answer on the report.
It is good for judging whether somebody can express an approach in code inside a timed sitting - a query, a loop, a fix, a signature. It is not a substitute for a full take-home where you need to know the program compiles and passes tests.
Any of 19 languages. The code is never run: AI reads it and marks it, so nothing has to be installed for any of them.
Pro and Business, alongside the rest of the AI grading. Marking is metered in AI credits; the attempt itself still costs the same single credit.
Not in the same paper. Coding questions go in a written paper, and a multiple choice paper takes multiple choice questions only. One assessment can combine both papers in one sitting, with the same proctoring on all of it.
Test the thinking, not the toolchain.
Short coding questions marked by AI, on Pro and Business. One credit for the sitting, AI credits for the marking.
- Free plan, 50 credits a month
- No install, runs in the browser
- Screen capture on every plan
- MCQs scored the moment they submit