Skip to content

Scan your resume,
free, on your own key.

Paste a job post and get a match report: the skills you’re missing, whether an applicant tracking system can read your file, and what to fix first. Then tailor the resume, track the application and prepare for the interview from the same self-hosted workbench. There are no pricing tiers and no monthly scan cap.

Free · self-hosted · bring your own Gemini key

Senior Backend Engineer · Northwind Payments

72%match rate · target 75%

Hard skills
31/45
Job title
15/15
Soft skills
6/10
Formatting
5/5

Hard skills

  • Kafka
  • PostgreSQL
  • Go
  • incident response
  • Missing: Kubernetes
  • Missing: gRPC

Searchability

  • Email and phone in the bodypassed
  • Standard section headingspassed
  • Two columns foundneeds attention
Example data. A real report adds recruiter tips, predicted skills and ATS tips.

Scan, read, fix, scan again.

The loop takes a couple of minutes per job, and rescanning a job you’ve already read doesn’t need your key.

  1. 1

    Add a resume and a job

    Pick a saved resume, upload a PDF or Word file, or paste text. Paste the job post, or import it from its link.

  2. 2

    Read the match report

    A match rate out of 100 with the breakdown behind it: skills found and missing, searchability, formatting and recruiter tips.

  3. 3

    Fix it, then rescan

    Edit beside the report with the score updating, or have Gemini tailor the resume line by line. Every scan is saved so you can see the number move.

What a match report tells you.

Gemini reads the job post once into a list of skills, a title, a level and an education requirement. Your resume is then scored against that list in code, so the same resume and job give the same number every time.

The weights are our own estimates and are listed in the breakdown, with hard skills carrying the most. The match rate is not any other tool’s number, and it can’t promise how a particular employer ranks you.

Hard skills
Each skill the job asks for, marked found, found by synonym (JS for JavaScript) or missing, with how much the posting stresses it.
Soft skills and other keywords
The interpersonal skills and industry terms the posting names, checked the same way.
Job title and education
Whether your titles line up with the role and your degree meets what the posting asks for.
Searchability
Name, email, phone and LinkedIn in the body of the document, a summary, standard section headings, consistent dates and a file type parsers accept.
Formatting
For uploaded PDF and Word files: tables, columns, text boxes, images, headers and footers, fonts and special characters that parsers trip over.
Recruiter tips
Measurable results, length, overused phrases, links to your web presence and whether your level matches the role.
Predicted skills
Skills the role usually needs that the posting never names, each with a one-line reason.
ATS tips
Paste the job link and the report names the applicant tracking system behind it, with guidance for that system.

Everything around the scanner.

Each of these is in the app today. They share one resume library and one saved job post, so a scan can lead straight into an edit, a tailored copy or a tracked application.

ATS identification

Import a job from its link and the report tells you which applicant tracking system it runs on. 25 systems are recognized, and each one's tips are drawn from the vendor's documentation or marked as general best practice.

Job link: boards.greenhouse.io/northwind/jobs/123

This posting runs on Greenhouse

Guidance for it is quoted from the vendor’s own documentation and cites its source.

  • Workday
  • Greenhouse
  • Lever
  • iCIMS
  • Taleo
  • SuccessFactors
  • SmartRecruiters
  • Ashby
  • Workable
  • Jobvite
  • and more

Power Edit

Edit your resume next to the report. Pick a job and the match rate and checklist recalculate as you type. Insert a missing skill or a saved bullet where your cursor is, and save the result as a new scan.

Match rate

7276

Kubernetes moved from missing to found. Example data.

Tailoring that shows its work.

Tailor a resume to a job and Gemini rewrites your real lines toward it. Each change arrives like this, with its reason. Nothing reaches your download until you’ve seen it, and nothing stays that you didn’t choose to keep.

Support Engineer, Larkspur Software

Handled customer Resolved Tier 2 escalations for enterprise accounts in Zendesk
Rewrite · Experience+4 pts

The posting asks for Tier 2 support experience twice. Your line describes that work without using the term.

Must-have2+ years of Tier 2 technical support

Verify“Zendesk” isn’t in your original resume. Keep it only if you used it.

Keep Revert
Example resume and job. In the app, the change list holds one card like this per change.
  1. A written reason

    Why the line changed, in plain words, pointing back at the posting.

  2. The requirement it serves

    The must-have or nice-to-have from the job that this change answers.

  3. A Verify flag

    Any new number, name or skill your original doesn’t back up is marked for you to confirm.

  4. Its score impact

    How many points this one change is worth, computed the same way every time.

  5. Keep or revert

    Your call, one change at a time. Reverting puts your original words back.

Resume builder, templates and sharing

Build a resume from scratch in eight guided steps. It comes out as an ordinary single-column document with standard headings. Choose from 5 templates, download PDF or Word, or share a read-only link, with an optional password.

Job tracker

Move each posting through Saved, Applied, Interviewing, Offer, Rejected and Withdrawn, with notes, contacts, interview dates and a follow-up date. Import and export CSV, and opt in to a daily email digest.

Browser clipper

A Chrome and Firefox extension saves the posting you're reading to your tracker. It connects with a personal API token and only reads a page when you open its popup. You build it from the repository and load it unpacked.

  1. 1Build the extension from the repository and load it in your browser
  2. 2Create a personal API token in Settings and paste it in
  3. 3Open a posting, review what it read, and save it

LinkedIn review

Paste the sections of your profile. See which of a role's keywords recruiters can find, what a complete profile still lacks, and get headline and About drafts built from your resume. Nothing connects to your LinkedIn account.

Keywords from the role, in your profile text

  • Kafka
  • Go
  • incident response
  • Missing: Kubernetes

Solid border: found. Dashed: missing. Example data.

Interview prep

Behavioral, technical and role questions for one job, each with what the interviewer is listening for. Type an answer, or dictate it where your browser supports speech input, and get feedback against the rubric.

The fixed rubric every answer is scored on

STAR structure
30
Specificity
25
Relevance
25
Measurable result
10
Length
10

Cover letter scanner

Measure a letter against a job: which requirements it mentions, whether it names the company and role, whether a claim is backed by a number, and which stock phrases to cut. You can also draft one for an application and edit it in place.

  • Names the company and the role
  • Backs a claim with a number
  • Mentions 3 of 5 requirements
  • Stock phrase found: team player

Example data.

Illustrations use example data.

Your server, your key, your data.

ResumeArchitect doesn’t resell AI and doesn’t meter you. You run it, you bring a Gemini key from Google AI Studio, and every request runs on your own account.

cp .env.example .env
# put openssl rand -base64 32 in KEY_ENCRYPTION_SECRET
docker compose up -d --build --wait
Setup in short. The README covers production, HTTPS and backups.
  • Runs on Docker Compose

    The web app, a background worker, Postgres, Redis and S3-compatible storage start with one command. An admin area covers users, queues, AI call logs, an audit log and announcements.

  • Your data stays on your server

    Resumes, job posts, scans and applications live in your Postgres database. Uploaded files go to your storage bucket. The production profile adds a nightly database dump, kept for 14 days.

  • Your key is encrypted at rest

    The Gemini key is stored with AES-256-GCM, under a secret that only your server has. It is decrypted on the server to make your requests.

  • Never shown again

    After you save it, the browser only ever sees the last four characters. You can replace or remove the key at any time.

  • What does leave the server

    When an AI step runs, the resume or job text it needs goes to Google under your key. Match scoring runs in your server's code and sends nothing out. Job links you import are fetched by your server.

  • No accounts to buy

    There are no plans, credits or usage tiers. Google bills you directly only if you choose a paid Gemini plan.

Questions, answered plainly.

The short answers on cost, limits, where your data lives and how your key is kept.

What does it cost? Is there a catch?
It is free, with no pricing tiers. You host it yourself and bring your own Gemini key, so the only money involved is what Google charges for your key, and its free tier works. If you choose a paid Gemini plan, Google bills you directly.
Is there a limit on scans?
There is no monthly or lifetime cap. To stop a runaway client from hammering your server, each account is rate-limited to 60 scans an hour, and to 20 tailoring runs an hour.
Where is my data stored?
On the server that runs this app. Resumes, job posts, scans and applications are rows in its Postgres database, and uploaded files sit in its S3-compatible storage bucket. Your records are private to your account. The text sent to Gemini when an AI step runs is the only thing that leaves.
How is my Gemini key stored?
It is tested against Google when you save it, then encrypted with AES-256-GCM under a secret held in your server's environment. The browser only ever sees its last four characters. It is decrypted on the server to make your requests, and you can replace or remove it in Settings at any time.
What does Gemini see?
Only what an AI step needs: the job post when it is read, an uploaded file when it is rebuilt into a resume, the lines being rewritten, or the text you give a career tool. Scoring a resume against a job you have already read sends nothing to Google.
How is the match rate calculated?
Gemini reads the job post once into skills, a title, a level and an education requirement. Your resume is then scored against that list in code, using weights we estimated that you can see in the breakdown. The same resume and job give the same number. It is not any other scanner's number.
Do I need a key to scan?
You need one to read a new job post. After that, scanning or rescanning a resume against that job works without it, and so does Power Edit.
Will the AI invent experience I don't have?
It is instructed not to, and you don't have to take that on trust. Every tailoring change is shown as a redline with its reason, and a separate check flags any number, name or skill your original resume doesn't back up as Verify. A missing requirement is only added when you say you have it.
Does a high score guarantee I'll get past an ATS?
No, and no tool honestly can. The report checks structure that parsers commonly struggle with and compares your wording with the job. How a particular employer's system ranks you is outside anyone's control.
Which model does it use, and what files can I use?
Gemini 2.5 Flash by default, which a self-hoster can change with the GEMINI_MODEL setting. You can scan or import PDF, Word (.docx) or plain text files (up to 8 MB to scan, 10 MB to import). Download PDF or Word, and a Word file you upload comes back with only the wording changed.

Run your first scan
in a couple of minutes.

Free to use with your own Gemini key, with no pricing tiers and no monthly scan cap. Your master resume is never changed by a scan or a tailoring run.