JobRadar

A real fit score, not just a sort-by-date list

Every listing gets a fit score from 0–100% instead of being dumped in chronological order. It's a deterministic, weighted keyword match — no randomness, fully explainable:

1. 1You define a skill profile in config/skills_profile.json — keywords with weights, so you can tell JobRadar which skills matter most.
2. 2Every scraped job description is checked against each keyword with a case-insensitive, word-boundary regex — so sql won't falsely match inside mysqlite, and multi-word skills like rest api are matched as a phrase.
3. 3The results are sorted by score, descending — the most relevant listings float to the top instead of being buried in an unsorted pile.

Change config/skills_profile.json and every score, and the ranking, updates to reflect it — no code changes needed:

{
  "profile_name": "Python Backend / Data Roles",
  "skills": [
    { "keyword": "python",     "weight": 3.0 },
    { "keyword": "django",     "weight": 2.0 },
    { "keyword": "rest api",   "weight": 2.0 },
    { "keyword": "docker",     "weight": 1.5 },
    { "keyword": "kubernetes", "weight": 1.0 }
  ]
}

Two commands, one ranked shortlist

Requires Python 3.8+. By default it parses a bundled local HTML fixture so the full scrape → score → dedup → export pipeline runs offline, deterministically, with no live network calls.

# install dependencies
$ pip install -r requirements.txt

# search + score + export
$ python jobradar.py --role "Python Developer" --location "Bangalore"