Offline AI Resume Screening System powered by Ollama + Qwen3:8B

QWENTUS is a fully local AI-powered Resume Screening System that automatically parses resumes, extracts public profile links, analyzes candidate information using a locally hosted Large Language Model (Qwen3:8B via Ollama), generates recruiter-ready Excel reports, and sorts resumes into decision folders — all without sending resume data to the cloud.

• Offline AI Resume Screening
• Local LLM using Ollama + Qwen3:8B
• PDF Text Extraction (PyMuPDF)
• OCR Fallback for Scanned/Image PDFs
• Hyperlink Detection
  • Visible URLs
  • Embedded PDF Hyperlinks
• GitHub Analysis
• LinkedIn Analysis
• Portfolio Analysis
• Resume Evaluation
• Automatic Resume Sorting
• Recruiter Excel Report
• Watch Mode
• Rich CLI
• Retry Logic
• Graceful Failure Handling
• Timing Metrics
• Production Logging

Incoming Resume
       │
       ▼
 Parse PDF
       │
       ▼
OCR (If Needed)
       │
       ▼
Extract Hyperlinks
       │
       ▼
GitHub
LinkedIn
Portfolio
       │
       ▼
Candidate Profile
       │
       ▼
Prompt Builder
       │
       ▼
Ollama
(Qwen3:8B)
       │
       ▼
JSON Evaluation
       │
       ▼
Excel Report
       │
       ▼
Move Resume

Add screenshots here.

Example:

• Startup Banner
• Resume Processing
• Excel Report
• Watch Mode
• Completion Summary

git clone <repository-url>
cd QWENTUS

python -m venv .venv

.venv\Scripts\activate

pip install -r requirements.txt

Download Ollama

https://ollama.com/

ollama pull qwen3:8b

ollama list

Single Run

python main.py

Watch Mode

python main.py --watch

Debug

python main.py --debug-one path/to/resume.pdf

Run Tests

python -m unittest discover -s tests -v

QWENTUS/

├── agents/
├── core/
├── exporters/
├── llm/
├── logs/
├── models/
├── parsers/
├── scrapers/
├── tests/
├── ui/
├── utils/

├── workspace/
│   ├── Incoming/
│   ├── Processing/
│   ├── Shortlisted/
│   ├── Maybe/
│   ├── Rejected/
│   ├── Failed/
│   └── Reports/

├── config.py
├── implementation.md
├── report.md
├── testingReports.md
├── main.py
└── requirements.txt

flowchart TD

A[Incoming PDFs]

--> B[PDF Parser]

B --> C[OCR]

B --> D[Hyperlink Extraction]

D --> E[GitHub]

D --> F[LinkedIn]

D --> G[Portfolio]

C --> H[Candidate Builder]

E --> H

F --> H

G --> H

H --> I[Prompt Builder]

I --> J[Ollama]

J --> K[Qwen3:8B]

K --> L[JSON Validation]

L --> M[Excel Report]

L --> N[Folder Sorting]

sequenceDiagram
    actor Recruiter

    participant Workspace
    participant PDFParser
    participant OCR
    participant LinkExtractor
    participant GitHubScraper
    participant LinkedInScraper
    participant PortfolioScraper
    participant CandidateBuilder
    participant Ollama
    participant ExcelExporter

    Recruiter->>Workspace: Drop Resume.pdf
    Workspace->>PDFParser: Parse PDF

    alt No selectable text
        PDFParser->>OCR: Run OCR
        OCR-->>PDFParser: Extracted text
    end

    PDFParser->>LinkExtractor: Extract hyperlinks

    LinkExtractor->>GitHubScraper: Fetch GitHub profile
    GitHubScraper-->>LinkExtractor: GitHub summary

    LinkExtractor->>LinkedInScraper: Fetch LinkedIn profile
    LinkedInScraper-->>LinkExtractor: LinkedIn summary

    LinkExtractor->>PortfolioScraper: Fetch Portfolio
    PortfolioScraper-->>LinkExtractor: Portfolio summary

    LinkExtractor->>CandidateBuilder: Build candidate profile
    CandidateBuilder->>Ollama: Generate evaluation prompt
    Ollama-->>CandidateBuilder: JSON evaluation

    CandidateBuilder->>ExcelExporter: Write candidate scores
    ExcelExporter-->>Workspace: Generate Excel report

    CandidateBuilder->>Workspace: Move PDF to Shortlisted / Maybe / Rejected / Failed

Incoming Resume
        │
        ▼
PDF Parsing
        │
        ▼
OCR (Optional)
        │
        ▼
Hyperlink Extraction
        │
        ▼
GitHub / LinkedIn / Portfolio
        │
        ▼
Candidate Builder
        │
        ▼
Prompt Builder
        │
        ▼
Ollama
        │
        ▼
Qwen3:8B
        │
        ▼
JSON Validation
        │
        ▼
Excel Generation
        │
        ▼
Resume Sorting

QWENTUS generates a recruiter-friendly Excel workbook.

• All Candidates
• Shortlisted
• Maybe
• Rejected
• Summary

Configuration is centralized in

config.py

Examples:

• OCR Engine
• LLM Timeout
• Retry Count
• DNS Servers
• Folder Paths
• Threshold Scores

Run Once

python main.py

Watch Mode

python main.py --watch

Debug

python main.py --debug-one resume.pdf

Tests

python -m unittest discover -s tests -v

Current Optimizations

• OCR only when required
• Embedded hyperlink extraction
• Prompt size optimization
• Retry logic
• Network health checks
• Graceful failure recovery
• Production logging
• Local inference
• Automatic Excel generation

1. Fork the repository

2. Create a feature branch

git checkout -b feature/my-feature

3. Commit changes

git commit -m "feat: add new feature"

4. Push

git push origin feature/my-feature

5. Open a Pull Request

Apache-2.0 License

Special thanks to the open-source community.

• Ollama
• Qwen Team
• PyMuPDF
• EasyOCR
• PyTorch
• BeautifulSoup4
• Requests
• OpenPyXL
• Watchdog
• Rich
• PyFiglet
• Pydantic
• orjson