Local Job Search Engine Scores Listings Against Your Resume
Job hunting is broken. Open LinkedIn. Scroll. Click. Read. Close the tab. Repeat, fifty times a day. There is no filter. No signal. No way to know if a listing deserves your next twenty minutes before you dive in.
That is exactly the problem Mohamed AlQouqa decided to fix. So he built a local job search engine that scores every listing against your resume, and it needs no API keys, no subscriptions, and no cloud access whatsoever.
The tool is called Job Tracker. It starts by parsing your resume automatically. Then it fires up to 31 search queries across platforms like LinkedIn, Indeed, Greenhouse, Lever, Ashby, and Himalayas. From there, it scores every result from 1 to 10 across five dimensions: title match, skill overlap, industry fit, salary, and location. The full stack runs on Node.js and Python, powered by python-jobspy. Importantly, your data never leaves your machine.
The scoring logic is clean and deliberate. Listings rated 7.5 to 10 earn the HIGH flag, strong matches across title, skills, and industry. Scores from 5.5 to 7.4 land in MED territory, worth a closer look. Anything below 5.5 drops to LOW, typically because the role sits in the wrong region, the salary falls short, or the industry is a poor fit.
Beyond that, the tool applies fuzzy title matching to keep scores honest. As a result, a Senior Project Manager listing will not score HIGH just because the word manager appears somewhere in a sales rep’s profile.
Building this local job search engine that scores listings against a resume brought real technical hurdles. On Windows, for instance, the system silently redirects the Python command through the Microsoft Store, returning no output and no error.
AlQouqa wrote a custom function to track down the real interpreter. He also built 12 semantic synonym groups for skills, so managed teams correctly matches team leadership. Furthermore, the tool graduates skill hits, more matches push the score higher without inflating it artificially.
The results from the first real run speak for themselves. The tool pulled 167 jobs across 31 queries and identified 8 HIGH matches in seconds. The top result was a Regional Sales Manager role at Hermès in Manhattan, scoring 8 out of 10. In total, AlQouqa estimates he saved around four hours of manual scrolling in a single session.
Looking ahead, he plans to add Mac and Linux support, a per-listing cover letter generator, and Claude API integration for AI-powered reasoning on each job card.
For now, the project lives on GitHub under an open licence. Windows users can download the zip, run setup.bat, and get started, no admin rights needed.
