Skip to content

Take-home assessments overview

A take-home assessment is an interview a candidate completes on their own time, in their own private workspace, instead of live with an interviewer. You invite a candidate by email, they solve your problem in a real sandboxed workspace on a timer, the assessment freezes when they submit, and you review the code, the session recording, and how they got there.

Where a live interview is a shared workspace you drive together in real time, a take-home is asynchronous: the candidate works alone, and you review afterward.

  1. Create the assessment. Start from a workspace, apply a problem (starter files, prompt, rubric), and turn on take-home mode. This workspace is the source assessment.
  2. Invite candidates by email. Each invite is private to one address. See Inviting candidates.
  3. The candidate accepts and starts. Accepting clones the assessment into a private workspace only that candidate and your reviewers can see. They click Start when ready, and the clock begins. See The candidate experience.
  4. They work, optionally with an AI assistant. If you enabled it, the candidate has a budget-capped AI assistant in the workspace, and every prompt is recorded for you.
  5. The assessment submits. A timed assessment auto-submits at the deadline; a due-date assessment is submitted by the candidate (late submissions are flagged). On submit the workspace freezes read-only, then Typelets autogrades in the background.
  6. You review and edit scores. Browse the submitted files, replay the recording, read the AI transcript, and change any autograded chip you disagree with. See Reviewing submissions and Grading. The source stays under Assessments until every accepted clone is submitted and graded, then it moves to Past assessments.

The assessment you build and invite from is the source. When a candidate accepts, Typelets makes them a private clone of it to work in. Reviewers never see those clones in the workspace switcher: the source is where you send invites, watch submission states, and review every candidate’s work. Candidates see only their own clone.

The workspace switcher groups take-homes the same way it groups live interviews.

  • Assessments - source assessments that still have work in flight. A candidate has not started, is in progress, or has submitted but not every accepted clone is graded yet.
  • Past assessments - muted, same pattern as Past interviews. A source moves here once every accepted clone is submitted and graded (autograde auto-apply or a human score). Opening one shows a Past assessment badge in the header.

Candidates see their own submitted clone under Past assessments as soon as they submit. That is earlier than the source moves for reviewers. See Reviewing submissions for the reviewer rule.

After sign-in, Candidate Dashboard is in the app menu - no reload. Hiring users use it to see candidates across take-homes. Candidates use it to find their own assessments. Once a grade is present, the dashboard treats that submission as scored; it does not stay on Awaiting review or Unscored. Candidates still never see the numbers. See The candidate experience.

Candidates see the prompt, their own files, and public test cases. They never see the rubric, hidden test cases (beyond pass/fail), the reference solution, other candidates, or any scores. See Sharing & roles.

Next: Inviting candidates.