I built a free tool that searches LinkedIn, Indeed, Glassdoor, ZipRecruiter and Google Jobs simultaneously

#python #apify #jobsearch #ai

The problem

If you've searched for a job recently, you know the drill.

Open LinkedIn. Search. Open Indeed. Search again. Open Glassdoor.
Search again. Switch tabs. Lose track of what you already applied
to. Miss jobs because you didn't check one board that day.

It's exhausting. And completely unnecessary.

What I built

JobHunt AI is a free tool that searches all major job boards
simultaneously in one run and returns clean, deduplicated results.

What it does:

Searches LinkedIn, Indeed, Glassdoor, ZipRecruiter and Google Jobs in one run
Supports multiple roles + multiple locations simultaneously
Auto-deduplicates results across all boards
Optionally scores each job against your resume using Claude AI
Returns a 0-100 match score, apply/consider/skip verdict, missing skills list, and top keywords per job
Exports to JSON, CSV, or Excel

How it works technically

Built on top of JobSpy
(MIT licensed Python library) wrapped as an Apify Actor.

JobSpy handles the multi-board scraping and proxy rotation.
Claude API (claude-sonnet-4) handles the AI resume scoring layer.

The architecture is deliberately simple:

scraper.py — JobSpy wrapper, board selection, dedup
scorer.py — Claude scoring, keyword extraction
main.py — Apify actor entry point, lifecycle
schema.py — Pydantic input/output models

That's it. No database. No auth. No unnecessary abstraction.

Example input

{
  "search_terms": ["AI engineer", "ML engineer", "LLM engineer"],
  "locations": ["Remote", "San Francisco", "New York"],
  "boards": ["linkedin", "indeed", "glassdoor"],
  "hours_old": 24,
  "results_wanted": 20
}

One run. Three roles. Three locations. All boards. Deduplicated.

Example output with AI scoring

{
  "title": "Senior AI Engineer",
  "company": "Example Labs",
  "location": "Remote",
  "source": "indeed",
  "ai_score": 86,
  "ai_verdict": "apply_now",
  "ai_missing_skills": ["Kubernetes"],
  "keywords": ["Python", "LLMs", "RAG", "evaluation"]
}