AI-Powered Resume Checker and Intelligent Filtering System

Streamlit-based project that parses resumes, compares them with a Job Description (JD), computes ATS scores, filters candidates, and explains selection/rejection with actionable insights.

Features

• Resume upload (.pdf, .docx) with automatic text extraction
• JD input (paste text or upload file)
• Stronger skill extraction and skill-gap detection
• Context-aware matching with TF-IDF cosine similarity
• ATS score with configurable weighted components
• Recruiter filtering (min ATS, min experience, required skills)
• Candidate ranking with decision explainability
• Resume quality analysis and improvement feedback
• Auto-generated interview questions
• Parsing warning when low text is extracted (common in scanned/image PDFs)
• Role-based authentication with login/register/logout (admin and customer)
• Advanced candidate search and view filters (status, shortlist-only, min probability)
• Analytics tab (status pie, ATS distribution, top skills)
• Comparison tab with side-by-side table and radar chart
• Saved Jobs manager (save/load/delete JD templates per user)
• Role Compare tab (score same candidate pool across multiple saved jobs)
• Admin panel for user management and global saved-job control
• Export both full results and filtered view CSV

Project Structure

resume_checker_ai/
|-- app.py
|-- requirements.txt
|-- README.md
|-- backend/
|   |-- __init__.py
|   |-- auth.py
|   |-- job_store.py
|   |-- parser.py
|   |-- jd_processor.py
|   |-- skill_extractor.py
|   |-- similarity_engine.py
|   |-- quality_analyzer.py
|   |-- scoring_engine.py
|   |-- filter_engine.py
|   |-- decision_explainer.py
|   `-- interview_generator.py
|-- utils/
|   |-- __init__.py
|   |-- constants.py
|   |-- helpers.py
|   `-- text_cleaner.py
|-- models/
|-- data/
`-- tests/

Architecture

Streamlit UI
   |
   v
Auth Layer (Login/Register)
   |
   v
Controller (app.py)
   |
   +--> Resume Parser
   +--> JD Processor
   +--> Skill Extractor + Similarity Engine
   +--> Scoring Engine
   +--> Filtering Engine
   +--> Decision Explainer + Interview Generator
   |
   v
Ranked Results + Insights Dashboard

ATS Scoring Formula

Default weights:

• Skill Match: 30%
• Semantic Similarity: 25%
• Experience Match: 15%
• Education Match: 10%
• Resume Quality: 20%

Final ATS = weighted sum of all component scores

Setup

1. Create and activate a virtual environment (recommended).
2. Install dependencies:

pip install -r requirements.txt

3. Run the app:

streamlit run app.py

How to Use

1. Provide job description in sidebar (paste/upload).
2. Optionally save the JD in Saved Jobs for reuse.
3. Upload multiple resumes.
4. Adjust filters and score weights.
5. Click Run Screening.
6. Review Candidates, Analytics, and Compare tabs.
7. Use Role Compare tab to compare candidate fit across saved job roles.

Admin vs Customer

• Customer account:
  • Full resume screening workflow
  • Saved Jobs (own account)
  • Candidate analytics, comparison, role comparison
• Admin account:
  • Admin control center
  • View all users
  • View all saved jobs across users
  • Delete any saved job

Push to GitHub

git init
git add .
git commit -m "Initial commit"
git branch -M main
git remote add origin https://github.com/<YOUR_USERNAME>/<YOUR_REPO>.git
git push -u origin main

Deploy (Streamlit Community Cloud)

1. Open https://share.streamlit.io
2. Login with GitHub and click New app
3. Select:
   • Repository: <YOUR_USERNAME>/<YOUR_REPO>
   • Branch: main
   • Main file path: app.py
4. Click Deploy

If build fails, check app logs and ensure requirements.txt installs correctly.

Notes

• Parsing quality depends on resume formatting quality.
• Scanned/image PDFs often have very low extractable text without OCR.
• Current semantic engine uses TF-IDF for lightweight local execution.
• You can replace similarity layer with sentence-transformer embeddings later without changing UI flow.