Skip to content
rankr
Sign inStart free

Take the exercise yourself.

The fastest way to judge Rankr is to sit on the candidate's side of it. Review the material, use the assistant, submit a verdict, and read the score the way an interviewer would.

What you will see.

  1. The material. A ticket and the pull request an AI opened for it, file by file, with the diff and the whole repository to browse.
  2. The assistant. Ask it anything about the change. It raises issues of its own; some are real, some are not, and accepting a wrong one costs points.
  3. The verdict. Leave comments on lines, then approve or request changes with a summary for the author.
  4. The score. Each planted flaw with whether you caught it, caught it with the assistant, or missed it, plus how you used the assistant.

Transaction-safe async TypeScript pitfalls

This is a Rapid fire exercise at senior level. A timed deck of short cards, each a snippet or a small diagram with one claim. True, false, or not sure, with a few seconds per card. It shows whether their confidence matches their accuracy when there is no time to check.

  • About twenty minutes, no time limit.
  • No assistant in this exercise.
  • Your work is scored at the end and you see the planted answer key, which a real candidate never sees.
  • Nothing is stored about you; the sample session is anonymous.

Needs a screen at least 1,100 pixels wide.