A simulation is a short, timed piece of real work. Before anything starts you see a preparation screen with the time limit, the attempts available, the skills the simulation verifies, and how your response will be evaluated. The timer starts only when you click Begin.
Where work samples run
Before your first work sample you answer one question, once: do you live in Manhattan, Brooklyn, Queens, the Bronx or Staten Island? Luminid cannot run work samples for people who live in New York City yet. If you answer Yes, no work sample opens, so there is no work sample score, and your applications stay with each hiring team. If you answered by mistake, write to hello@luminid.org.
The formats
Simulations use one or more of these response formats.
- Written answers you type in the browser
- Code you write in a plain editor in the browser
- Items you drag into order of priority
- Questions where you pick the best option
- A spoken answer you record in the browser
- A live call you hold in the browser with an AI counterpart
For the two voice formats you need a working microphone and a quiet spot. Voice responses are transcribed and the transcript is what gets scored. Scoring looks at the substance of what you say, never your accent or fluency. If a spoken response is not accessible to you, you can request an accommodation from the start screen before you begin.
Some simulations have several parts, and some end with a short reflection where you write what you would approach differently. Save each part as you go.
Live voice calls
Before a live call you check your speaker and your microphone on the preparation screen. You play two short tones and confirm you heard them, then say a few words until the meter hears your voice. Begin stays locked until both pass, so a muted speaker or a blocked microphone is caught before the call, not during it.
Saving the call recording is on by default. You can turn it off, and you will be asked to confirm first, because hiring teams listen to recordings and some treat a missing one as a reason not to move forward. With the recording off, only the transcript is kept.
During the call there is one button, End call and submit. Ending the call sends it. If the call fails on our side, you get the attempt back automatically. What you said is saved as you go, so a dropped line does not lose the call.
What is allowed
The work must be yours, in your own words. Two things run in the background, and we tell you about both before you start.
- Browser automation is blocked. If automation tooling is detected controlling the page, the session stops. Verifications must be completed by you.
- The session records technical integrity signals, such as typing and cursor patterns, to detect automated activity. These signals protect integrity only. They are never shown to recruiters and never affect your score.
Switching tabs is recorded as a count, nothing more. There is no penalty, no lost attempt, and no auto-submit. Real life includes notifications, translation tools, and interruptions, so the count is shown to recruiters as plain context and carries no judgment.
After you submit
Scoring usually takes 10 to 15 seconds. Your results page shows each skill separately, with what you did well, areas to grow, and the exact quotes from your response that support each result. When the work sample can add a verified skill, and its start screen tells you, skills that meet the bar are saved to your profile as verified. You decide what to share from there.
When the simulation is a company's hiring assessment, you get a receipt instead, on screen and by email. It names the company that has your call, when it arrived, and whether the recording or only the transcript went with it. The company contacts you directly about the next step.