AI Policy

ResumeRama is an AI product. That means your resume text is sent to a third-party AI provider to be rewritten. This page explains precisely what is sent, what is removed first, and what we will not do.

Which providers we use

Production runs on Microsoft Azure AI, using the Azure-hosted DeepSeek-V4-Pro model, in the United States (Azure East US). An Azure-hosted OpenAI model is configured in the same region as a manual alternative.

Two things follow from that, and both are the reason we name it rather than saying “our AI partners”. Your resume is not sent to DeepSeek the company or to any service in China - Azure runs the model weights under Microsoft’s terms. And automatic failover between providers is switched off, so a busy moment cannot quietly route your resume somewhere with different terms or a different country.

This matters to you, so we will publish it plainly rather than leaving it as “our AI partners”. Where the model runs determines which country your resume data travels to.

What we send

Only what the job needs: your resume content and the job description you are targeting. Specifically:

• Your resume - summary, experience, education, skills and the other sections you filled in
• Your contact block, as parsed from your resume
• The job description text
• Any special instructions you typed

We do not send your payment details, your password, your browsing behaviour, or your tailoring history.

What we remove before sending

Your contact block is not sent at all. Name, email, phone, location and your profile links are stripped before the request is built. The model is asked to rewrite your summary, experience and skills; it is never asked for contact details and has no field to return them in. Your finished document is assembled here, from your original resume.

Everywhere else we send - summary, experience, education, projects, the job description, your special instructions - we replace your name, email addresses, phone numbers, street addresses and URLs with placeholders. The provider sees [[PII_EMAIL_1]] rather than your address. When the response returns, our servers substitute the real values back.

Your name is matched by value, not guessed by pattern. We know it from your resume, so we can remove it reliably wherever it appears, including inside a bullet point. Redaction walks every field of the document rather than a chosen list, so a field added later is covered the day it is added.

An earlier version of this page said the provider “never receives enough to identify you” while the contact block was in fact being sent in full. That was corrected rather than left standing, and the engineering gap behind it has now been closed - which is why this page can state the stronger position again.

Where processing happens

Redaction and rehydration both run on our servers in Canada, not in your browser. Your data is stored in Canada; the AI call is the point at which it may leave. See International transfers in the Privacy Policy.

Training

We do not use your resumes, job descriptions or account data to train AI models. Where we use models hosted on Microsoft Azure, the terms prohibit training on our traffic. For providers called directly, TODO: confirm - training terms per provider, and we will state the answer here rather than imply one.

If we ever build a programme to improve models using your data, it will be strictly opt-in: off unless you deliberately turn it on, explained in plain language beforehand, and reversible at any time. Anything created before you opt in stays out of training entirely.

Treat the output as a draft

Language models produce fluent text that can be confidently wrong. They can overstate experience, invent detail, or rephrase something into a claim you cannot support. You are responsible for the accuracy and honesty of your final resume - read every line before you send it. Match scores are estimates to guide edits, not predictions of any employer’s system. More in the Disclaimer.

What AI does not decide

ResumeRama does not screen, rank or reject you, and does not send results to employers. Its output is a document you edit and choose what to do with. If you want a human to explain how a result was produced, ask us.

Limits on use

Do not use ResumeRama’s AI features to generate deliberately false credentials, to impersonate someone, or to process another person’s resume without their knowledge. See the Site Usage Policy.

Changes

If we change providers, what we send, or how redaction works, we will update this page, date the change at the top, and ask you to re-accept where the change is material.