YOUR PRACTICE IS PERSONAL

Privacy at TensorReps

Effective October 19, 2026. This notice describes the current TensorReps beta.

What we collect

When you sign in with Google, we receive your basic identity, including your name, email, and a provider identifier. We create a stable TensorReps user ID. We do not request access to your Gmail inbox or store your Google password.

We store your profile, code drafts, submissions, execution results, bookmarks, notes, and feedback. Optional coaching saves messages and transcripts, and saves microphone and coach audio only if you select private audio playback. Interview sessions also save code revisions, messages, live voice transcripts, requested hints, run evidence, deadlines, and debriefs for private replay. Interview voice audio is processed live but is not saved as a recording. Signed-out drafts stay in your browser tab. For recovery, a signed-in draft may also be stored temporarily in that browser tab under your user ID.

How we use it

We use this information to sign you in, restore your work, run code you submit, show your progress, investigate problems, and limit abuse. Your profile, code, submissions, and notes are private in this beta. We do not sell your personal information or use advertising trackers.

Services that process data

Google handles sign-in. Neon stores account and practice data and provides authentication. Vercel hosts the application and may process request metadata such as IP addresses and operational logs. When execution is enabled, E2B receives the code you choose to run inside a temporary isolated environment. If you choose AI coaching or an AI interview, OpenAI processes that session’s exercise, code, messages and run evidence to provide interviewer responses and process feedback. Text Responses API requests use storage disabled; realtime voice is processed during your connection and provider abuse-monitoring retention may still apply. Do not submit passwords, API keys, personal datasets, or confidential employer code.

Cookies and security

Essential cookies maintain your login session. We use a one-way keyed hash of an IP address to apply execution limits; this hash is not shown to other users. Code runs have no application database credentials. We request microphone access only when you enable voice in coaching or an AI interview. Stopping voice or leaving the workspace disconnects the microphone. Interview voice also ends when the round finishes or its deadline is reached. We never request camera access. Optional recordings are held in private Vercel object storage, accessible only through your authenticated account.

Protecting the private library

We record the identifiers of questions and solutions opened, request counts, pseudonymous session identifiers, and access-limit decisions to prevent bulk collection and account sharing. Content and security records are retained for up to 30 days; owner access-decision records for up to 90 days. The owner can review these records and suspend access. A pseudonymous member mark may appear on study content. Vercel BotID processes browser signals for verification; these checks do not grant additional content access.

Retention and control

You can export your practice data and delete your account in Settings. Deletion removes active account and practice records. Limited operational logs, abuse-prevention counters, and provider backups may remain for their retention periods. An anonymized execution reservation may be kept to enforce a shared monthly usage allowance.

Feedback is retained for product improvement; the user association is removed when an account is deleted. Avoid including personal information in feedback you do not want retained.

Questions or requests

Use the feedback form on the About page for privacy questions or requests. We will update this notice as the service changes.