A smart resume screening tool that actually reads resumes like a human would. Upload a resume, paste a job description, and get a detailed match score with real insights.

Takes the boring out of resume screening. The AI reads through resumes, pulls out what matters, and tells you if someone's a good fit for your job opening.

• Catches hidden skills – Finds relevant skills mentioned in project descriptions that candidates forgot to list in their skills section

• Structured data extraction – Uses Instructor to parse resumes into clean, organized data instead of messy text dumps

• Remove catch phrases – Remove corporate jargon used to nump up resume scrore

• Detailed scoring – Get scores for skills match, experience level, education fit, projects, certifications, and even cultural fit

• Custom weights – Adjust what matters most to you. Need someone with tons of experience? Bump up that weight. Looking for fresh ideas? Focus on projects

• Lightning fast – Powered by Groq for inference that doesn't make you wait

• Flexible fields – All fields are optional. Got a resume without a portfolio? No problem. Missing education? We'll work with what we have

• Smart caching – Same resume uploaded twice? We remember it. Only re-screens if the job description changes. Saves time and API costs

cd backend
pip install -r requirements.txt

Create a .env file with your API keys:

GROQ_API_KEY=your_groq_key_here
SUPABASE_URL=your_supabase_url
SUPABASE_KEY=your_supabase_key

Optional: Set up caching (recommended!)

• See backend/SUPABASE_SETUP.md for Supabase cache setup
• Without caching, the app still works but will be slower and more expensive

Run the server:

python app.py

Just open frontend/index.html in a browser. That's it.

POST /api/parse – Just parse a resume, no screening

• Upload a PDF or DOCX file
• Get back structured data (name, email, skills, work history, etc.)

POST /api/screen – Full screening against a job description

• Upload resume + job title + job description
• Get detailed match analysis with scores and recommendations

• Groq – Fast LLM inference
• Instructor – Structured LLM outputs with Pydantic
• Flask – Simple API server
• Python-docx – Word doc parsing
• Vanilla HTML/CSS/JS for the frontend

Want to see what the actual output looks like? Check these out:

• Parsed resume example – See how the parser extracts structured data from a resume
• Screening result example – Full screening analysis with scores and recommendations

{
  "overall_score": 7.83,
  "recommendation": "Potential Match",
  "skill_match": {
    "score": 8.5,
    "matched_skills": ["Python", "Java", "JavaScript", "CI/CD"],
    "missing_skills": ["Direct experience with AWS, GCP, or Azure"],
    "additional_skills": ["Machine Learning", "NLP", "Computer Vision"]
  },
  "strengths": ["Strong technical skills", "Impressive project portfolio"],
  "concerns": ["Limited experience in software development"]
}

• PDF files (.pdf)
• Microsoft Word (.docx, .doc)

Do whatever you want with it.