Free forever. Runs on your laptop.
Job hunting is a full time job. Nobody pays you for it.
So this does the boring half. It finds the roles, scores them against what you have actually done, and writes the resume.
Why this exists
Final year, off campus, and the whole thing was manual. Open twenty career pages, read the same job description written five different ways, guess whether I stood a chance, then rewrite my resume for the fourth time that week.
The part that got me was how little of it was thinking. It was tab management. So I spent my evenings building the thing I wanted to exist, and then I used it to apply to the jobs it found.
It is open source because there is no reason it should not be.
What it does
Reads careers pages straight from ninety plus companies on Greenhouse, Lever, Ashby and SmartRecruiters. Search any company by name to add it, or paste a job description you found somewhere else.
No feed, no promoted posts, no listings that closed in March.
Pick your region and it only keeps jobs you could actually take. Roles asking for more years than you have, senior titles and the wrong function are dropped before you ever see them.
Every exclusion carries the rule that caused it, in plain words.
Every role scores out of a hundred from five weighted parts: required skills, meaning, preferred skills, keywords and your stated preferences. No model is asked whether it likes you.
Open any score and read the five numbers that made it.
Every bullet is built from something already in your knowledge base and traced back to it. Anything the model cannot trace, and any number it invents, is thrown away before the page is made.
One page, ATS readable, checked by reading the PDF back out again.
Paste the LaTeX resume you already spent a weekend on and it rewrites the wording section by section, in your layout. A section it cannot rewrite truthfully is left exactly as you wrote it.
Opens straight into Overleaf when you are done.
Builds targeted searches for the five people worth contacting at a company, alumni first, with a short opening message grounded in what you have actually built. Then it suggests one project to build for that role.
Nothing is scraped. You run the search, so your account is never at risk.
Your laptop
No account, nothing to log into, no API key to buy. The model runs locally through Ollama and everything it knows about you is one file on your own disk. The only requests it makes are to public job boards, which is what happens when you open a careers page yourself.
There is no server bill, because there is no server.
Your career history is one file that never leaves your laptop.
Back it up by copying it. Read it, change it, delete it.
Before you start
This is not a search engine you type into. It reads what you have actually done and matches it against real postings, which means the half hour you spend feeding it is the half hour that decides whether the output is any good.
Read the guideEvery project, internship, hackathon, certification. In detail, with numbers.
It decides which jobs are even stored, not just which ones are shown.
Open the shortlist, generate, send. The long part happens once.
Questions
Free, and there is nothing to charge for. It runs on your laptop, so there is no server to pay for and no usage to meter. There is no account, no plan, and no card. The code is public and you can read exactly what it does.
Nowhere. Your career history is one file on your own disk, and the model that reads it runs on your machine through Ollama. The only requests the app makes are to public job boards, which is the same thing that happens when you open a careers page yourself.
You need to be willing to copy four commands into a terminal once. The page walks you through opening one. After that it is an ordinary website in your browser. If that first part sounds like too much, this is honestly not ready for you yet.
A laptop with a few GB free. The setup script installs Python, Node and Ollama itself if they are missing, then downloads a model that is about 2 GB, once. There is no database to install and no Docker.
About fifteen minutes, and nearly all of it is a download you can walk away from. Starting it again on another day takes two commands and a few seconds.
No, and that is deliberate. It finds the roles, scores them, and writes the resume. You read it, you decide, and you press send on the company's own page. Nothing is submitted anywhere on your behalf.
It is built so that it cannot. Every bullet has to cite the accomplishment it came from, and any number in a bullet has to already appear in that accomplishment. A line that fails either check is dropped rather than shipped.
It reads public job boards, so coverage follows the companies that use them. You set your region first, and that decides which roles are stored at all. India, the US, the UK, Europe, Canada, Australia and Singapore each have their own list, and there is an option for no location filter.
Yes. If you keep your resume in Overleaf, you can hand it your LaTeX and it rewrites your own document rather than handing you someone else's layout. There is also a plain PDF built for machine readers if you do not have a template you care about.
No. The wall is a headcount of people doing the same thing, and joining it opens the setup steps on this page. The tool itself does not know or care whether you joined.
Getting it running
About fifteen minutes, and most of that is a download you can walk away from. No database to install and no Docker. You do not need to have anything set up first: the script installs Python, Node and Ollama itself if they are missing.
2595 people on the wall