What happens to what you share

Camera and microphone

• The camera self-view stays on this device. It is never recorded, uploaded or sent to a model.
• Your microphone audio is streamed through our server to Meta's Muse Voice Transcribe, which turns it into text. Audio is never recorded or saved; only the text transcript is kept, as part of your answers below.
• The interviewer's voice is your browser's speech synthesis, running on your device.
• Natural voice (beta) is not offered right now while its audio is fixed, so nothing is downloaded for it. When it is offered again, it will be only in browsers with WebGPU. If you turn it on, the interviewer's voice is made by the Kokoro model running in your browser. Its text never leaves your browser; your browser only downloads the model from Hugging Face the first time.
• Accent, delivery and pace are never scored.

Your answers

• Answer text is kept only with your consent: the checkbox before you start.
• Your answer text is sent to a model (Muse Spark 1.3 from Meta, or Muse Spark Open hosted on Baseten) to find the number you committed to and your reasoning. Code, not the model, checks the number.
• On our server, your answers are kept in memory for this interview's report, which is gone after a restart. They are not written to disk.
• In coding rounds, your Python runs in your browser, not on our server. Your code and its results are sent to our server and the model so the interviewer can ask about them.
• Exporting an interview happens only when you choose to, and the file is saved to your device.

Bring your own question

• An image you upload is sent to the model to read the question, then dropped. It is not stored or logged; we keep only a fingerprint of it (a SHA-256 hash).
• The question text, and the code models write to check it, are kept in memory for your session only and are not written to disk; they are gone after a restart. That code runs in a sandbox on our server and is never added to the question bank.

Your profile and account

• Without an account, your profile stays in this browser. Nothing about it is stored on our server.
• Signing in is optional (Google, or an email link) and uses Firebase from Google. Firebase keeps your sign-in: your email address, plus your name and photo if you use Google.
• Signed in, your profile is saved to your account: your roles, display name, photo, ratings, per-answer scores, interview summaries and placement. Answer text is never stored there. Only you can read your account.
• You can delete your account on the Profile page. That removes your profile, ratings and history from our database and this browser.
• "Share this so we know what to build next" sends only the roles you picked and your optional note: no account, email or IP address.

Limits and your IP address

• To keep model costs in check, the number of interviews, question checks and transcriptions each visitor can start is limited. Your IP address is used only as the key for these counters. They are kept in memory, never logged or stored, and cleared when the server restarts.

Other services this site uses

• Meta (Muse Spark 1.3 and Muse Voice Transcribe) and Baseten (Muse Spark Open): the models.
• Firebase (Google): optional sign-in and accounts.
• Google Fonts: the page fonts.
• Hugging Face and jsDelivr: the natural voice model and its library, only if you turn Natural voice on.
• cdnjs and jsDelivr: the code editor and Python for coding rounds, loaded only in a coding round.
• Render: hosting.

The services your browser loads directly (Render, Firebase, Google Fonts, cdnjs, jsDelivr and Hugging Face) see your IP address, as any website does. The models are reached only from our server. The site has no ads and no analytics.