Privacy Policy

1. What we collect, and why

We collect only what the product needs. Each purpose is specific; we don’t reuse data for unrelated purposes (purpose limitation). Section 5 says where each kind of processing happens.

We do not collect data we don’t need, and we do not buy data about you from anyone.

Application data is used only by the employer for that one opening, for as long as the application is active. Withdrawing an application revokes the employer’s access immediately, but the application record itself is retained until you delete your account or withdraw the application, which is what withdrawal removes access to, not the record. Directory data is used only to show your card to recruiters at claimed employer pages, and stops being shown the moment you turn the directory off.

1b. Sensitive personal data

Under the SPDI Rules, 2011 some of what Roleora holds is sensitive personal data: your password (we keep only a one-way hash), financial information in a document you upload (for example bank details on a salary slip), and any health information you choose to write in an entry or the journal. Passport number, address and reservation category in your application profile are encrypted the same way, though the law does not list them.

We collect this only with your consent, given when you sign up and again each time you add or upload it. You can choose not to provide it, and you can delete it at any time. We never sell it, never send it to an AI model unless you upload a document for reading, and share it only as Section 5 describes or when the law requires.

2. Consent, and withdrawing it

We process your personal data on the basis of your consent, given when you sign up and when you turn on optional features, after notice in clear, plain language. You can:

• withdraw consent anytime, as easily as you gave it, by turning off the relevant feature, or by deleting your account from Settings;
• request this notice and consent information in English (the language Roleora launches in) or in any language listed in the Eighth Schedule to the Constitution of India: write to privacy@roleora.com and we will provide a translation;

• withdraw your application anytime from your application tracker. Withdrawing revokes the employer’s access to the shared data immediately, the same revoke used for any other share link;
• turn off the employer directory anytime from Settings, Privacy. Your directory card is removed immediately and recruiters lose access to it going forward. This does not affect any application you have already submitted, which is a separate consent.

Withdrawing consent stops future processing; it doesn’t make past lawful processing unlawful, and some data may be retained where law requires (Section 7).

3. Our three commitments

1. No AI training on your data. Your content is never used to train or fine-tune AI models, not by us, and contractually not by our AI processor.
2. No selling data. We do not sell, rent, or trade personal data. No “anonymized insights” sold sideways either.
3. No ads. No advertising on Roleora. We use one analytics tool, Microsoft Clarity, and it runs only if you accept it. Microsoft uses what it collects under its own terms. See Section 11c.

We charge money for the product so we never have to make money from your data.

4. The encrypted journal: a special note

If you turn on journal encryption, your journal entries are encrypted on your device (AES-256-GCM) with a passphrase only you know. We store ciphertext. We cannot read, recover, or hand over your encrypted journal content, not to you without your passphrase, and not to anyone else. If you lose both the passphrase and your recovery code, that content is permanently unrecoverable. That’s the honest cost of a real lock. An encrypted entry is never translated or read by an AI model.

5. Who processes data on our behalf, and where

We use a small set of processors, each bound by contract to process data only on our instructions. Most of Roleora runs in India; three kinds of AI processing run outside it, named below.

• Microsoft Azure, Central India: the application, the database, your files, our logs, speech recognition and the voices of the practice interviewer (Azure AI Speech), and the sandbox that runs Python code in a coding round. The app’s web pages are served by Azure Static Web Apps from its East Asia region.
• Azure OpenAI Service (Microsoft), Sweden Central (EU): writing documents, chat answers, interview questions and feedback, reading your entries for skills, the English version of Hindi entries, reading uploaded marksheets, and the Community safety check. Under Azure OpenAI terms, your prompts and outputs are not used to train models and are not shared with other customers or with OpenAI.
• Azure OpenAI realtime models (Microsoft), South India and Sweden Central: the live spoken interview. Your voice is streamed to the model while you speak and is not stored by us. The same no-training terms apply.
• Azure AI Vision (Microsoft), East US: reading the text in scanned PDFs you upload, so a scanned resume can be imported.
• Azure Container Apps sessions (Microsoft), Central India and South India: running the code you write in a coding round, in an isolated sandbox with no internet access that is cleared minutes after use.
• Application Insights (Microsoft), Central India: error reports and performance traces from our servers, used to find and fix faults.
• Azure Communication Services (Microsoft): sending OTP and service emails.
• Razorpay: payment processing (an RBI-regulated payment aggregator). Razorpay handles your payment instrument under its own privacy policy; we receive only transaction status and references.
• Google: sign-in only. Google tells us who you are; we send it nothing from your record.
• GitHub: sign-in, and the optional connector that reads your own GitHub activity into a review inbox.

Some services get a small, fixed piece of data for one job and nothing else. When you set a password, we send the first five characters of its one-way hash to Have I Been Pwned to check it has not leaked; the password itself never leaves us. A job search you run sends your search words and locations to our job-search provider, JSearch (through RapidAPI).

Connectors are different. If you switch one on, we send that service only the handle, link or key you give us, and we use it to read your own work. The connectors open to you are listed in the app under Settings, Connections.

AI tools you connect, and the browser extension. You can connect an AI tool you use (through our MCP server) or the Roleora browser extension. You choose what it may do on a permission screen: by default, log work to your review inbox, read your entries, and read a profile summary. Your application profile is a separate permission, and its sensitive fields (such as passport number and address) are a further separate permission, asked on a second screen, covering only the fields you chose to share. Once a third-party AI tool has read your data, it holds that data under its own terms, not ours. The extension fills only the field you click into and never submits a form. Every read is listed under Settings, Connections, where you can disconnect at any time.

We use no advertising processors. Microsoft Clarity, our one analytics tool, is not our processor: see Section 11c.

5b. AI in Roleora, and how we label it

• The practice interviewer is an AI. It is a drawn character with a computer-generated voice. The interview room labels it as an AI interviewer, and its scores are AI practice estimates from one session.
• Documents are AI drafts. Resumes, letters and reports are written by an AI model from your own entries. For a resume, Roleora checks every number and every cited source against your record and marks the bullets it cannot match. It cannot catch every overstatement, so read before you send. A document you download is yours and carries no AI label from us.
• English versions of Hindi entries are AI translations. The app marks them so you can check them.
• Community moderation is AI-first, with a person for the final word. Every new post, comment and message is checked automatically. Keyword rules always run, and an AI classifier also reads it up to a daily limit; past that limit only the keyword rules run. When the check is highly confident that something is in a serious category (intimate images, sexual content, credible threats, doxxing or other people’s personal data, hate, encouraging self-harm, and money scams), it is hidden at once until a person reviews it. Two things are never hidden automatically: a post from someone who seems to be in distress or at risk of harming themselves, which goes straight to a person, and anything tied to a court or government order, which a person always handles. Other serious or uncertain cases also go to a person. Nothing is removed for good unless a person decides. To ask for a review, email grievance@roleora.com. Your posts are never used in your resume, your chat or your AI memory, and never used to train models.
• No automated decision about you. No AI model decides anything about your account, your access or your plan. The one automatic action is hiding Community content the check is highly confident is in a serious category, until a person reviews it.

6. Your rights (Data Principal rights under the DPDP Act)

You have the right to the following. We honour them now, before the DPDP Act makes them law on 13 May 2027.

• Access: a summary of the personal data we hold about you and how it is processed. Most of this is simply visible in the app; the rest, on request.
• Correction and updating: fix inaccurate or outdated data, mostly self-serve in the app.
• Erasure: delete individual items anytime, or delete your whole account from Settings. Deletion is real (see Section 7).
• Portability: a one-click JSON export of everything you own, free, anytime, from Settings.
• Grievance redressal: see Section 9.
• Nominate a person to exercise your rights if you die or are incapacitated (write to privacy@roleora.com to set this up).

To exercise any right the app doesn’t cover self-serve, email privacy@roleora.com. We verify identity via your registered email before acting on requests.

7. Retention and deletion

• Active accounts: we keep your data as long as your account is in use. It’s your record; that’s the product.
• Account deletion: immediate, irreversible erasure of your record from live systems on confirmation; backup copies expire within their 30-day retention window. Payment and tax records are kept as long as law requires, then deleted.
• What the law makes us keep for 180 days: the IT Rules, 2021 require two narrow holds. First, your registration details (account id, name, email, the dates your account was opened and deleted, and your sign-in times and IP addresses) for 180 days after the account is deleted. Second, any Community content we removed, or that was reported, with its report and moderation record, for 180 days after removal, even if its author deletes their account. Both are sealed, readable only by the Grievance Officer, used only to answer a court order or a lawful request from an authorised agency, and then deleted.
• Security logs: kept for 180 days, in India, as the CERT-In Directions of 2022 require, then deleted.
• Dormant accounts: after 24 months with zero logins, we send four notice emails, at 24, 25, 26 and 27 months, each with a one-click export offer. The last one names a deletion date at least 30 days away. A login or a click on keep stops it. If you don’t return, the account and data are deleted. Paying members are never treated as dormant. We don’t keep personal data nobody is using.

8. Children’s data

Roleora is for users 18 and older. We ask for your date of birth before you can use the app, and we refuse anyone under 18. We do not knowingly process a child’s personal data. If we learn an account belongs to someone under 18, we delete the account and its data. We do not do targeted advertising for anyone, of any age; features that observe your activity over time (streaks, heatmaps, AI memory) are the reason we restrict signup to adults, since the DPDP Act prohibits behavioural monitoring of children.

9. Grievance redressal (Grievance Officer)

Under the Information Technology Act and its rules, and the DPDP Act, our Grievance Officer is:

Aaryash Kansal

Director, KNETT IT INFRASTRUCTURE AND SOLUTIONS PRIVATE LIMITED

grievance@roleora.com

79, Shyam Lal Road, Darya Ganj, New Delhi, Delhi, India – 110002

We acknowledge a privacy grievance within 24 hours and resolve it within one month, as the SPDI Rules, 2011 require. Complaints about Community content follow the shorter times on the Grievance Redressal page. If you are unsatisfied with our response, you may approach the adjudicating officer under section 46 of the Information Technology Act, 2000 (the IT Secretary of your State), and from 13 May 2027 the Data Protection Board of India.

10. Data breaches

If a security incident affects personal data, we report it to CERT-In within 6 hours of noticing it, as Indian law requires, and we tell you without delay, in plain language: what happened, what data was involved, and what we are doing about it. From 13 May 2027 we will also notify the Data Protection Board of India as the DPDP Act requires. No burying it in a changelog.

11. Security

Encryption in transit and at rest; field-by-field encryption of the sensitive fields in your application profile; tenant isolation enforced on every database query; short-lived signed URLs for file access; malware scanning of uploaded files; rate limiting, audit logging, and security headers; two-factor sign-in; client-side encryption available for the journal. Payments never touch our servers.

11b. Cookies

Roleora sets first-party cookies that are strictly necessary to sign you in and keep your session secure. We set no advertising cookies.

Analytics cookies are separate, and they are off until you accept them. If you accept, Microsoft Clarity sets two first-party cookies: _clck keeps a random visitor id, and _clsk joins your page views into one session. If you decline, Clarity does not load and sets nothing.

Microsoft also lists third-party cookies for Clarity: CLID, ANONCHK, MR, MUID and SM, set from Microsoft domains. We always tell Clarity that ad storage is denied, which stops MUID. We cannot see or control the rest, so we state them here.

11c. Product analytics

On roleora.com and in the Roleora app we use Microsoft Clarity for product analytics: heatmaps, aggregated usage and session replays. We use it to find pages and screens that confuse people.

It runs only after you accept it. The banner has two buttons, Accept analytics and Decline. Declining is the default, it is remembered in your browser, and it loads nothing.

What it records, once you accept:

• Which pages and screens you open, where you click and scroll, mouse movement and how long you stay.
• Device, browser, operating system, approximate location (country or region) and where you came from.
• A random visitor id and session id, held in the two cookies above.

What it does not record:

• Text. Every piece of text on the page is masked in recordings, so a replay shows shapes, not words. Nothing you type is captured.
• Your record. Résumé, journal, plan and application content is masked like all other text.
• Your voice. The practice interview room and the journal do not load Clarity at all.
• Your name or email. In the app, Clarity gets a one-way hash of your account id and your plan, free or pro.

Microsoft is not our processor for Clarity. Under the Clarity terms, Microsoft and Roleora each decide separately how to use this data. Microsoft may also use it for its own purposes, including Microsoft Advertising. That use is governed by the Microsoft Privacy Statement. We tell Clarity that ad storage is denied. We do not sell this data or use it for advertising.

To withdraw, choose Analytics in the footer of any page, then Decline. In the app, turn off the Analytics switch in Settings. Clarity clears its cookies, the page reloads, and nothing loads after that.

12. Changes to this policy

Material changes are announced by email and in-app notice at least 15 days before they take effect for existing members, with the previous version archived. We will never change the three commitments in Section 3 quietly.

Earlier version: Privacy Policy as published on 4 October 2026.

Questions: privacy@roleora.com.