A powerful, privacy-focused Resume Parser that uses Local LLMs (Large Language Models) to extract information and generate insights from Resumes/CVs.

New: Now fully migrated from pyresparser to Agentic AI using Qwen2.5-1.5B!

• AI Power: Uses Qwen2.5-1.5B-Instruct (GGUF) for high-speed, accurate extraction.
• Smart Extraction: Extracts Name, Email, Mobile, Skills, Education, Experience, Companies, and more.
• Intelligent Insights: Automatically generates a Professional Summary and identifies Key Strengths for each candidate.
• Structured Data: Outputs strict JSON format for easy integration.
• Zero API Cost: Runs 100% locally. No OpenAI/Anthropic/Gemini keys required.
• Cross-Platform: Windows, macOS (Intel/Silicon), Linux.
• Portable: Can be built into a standalone executable.

• Model: Qwen2.5-1.5B-Instruct (Quantized: q4_k_m)
  • Why Qwen? It is significantly faster and more accurate for structured data extraction than many 7B models, while being lightweight (~1GB).
• Inference Engine: llama-cpp-python (with Metal/CUDA support).
• Context Window: 2048 tokens (input text is truncated to ~1500 chars to fit).
• Hardware Requirements:
  • RAM: Minimum 4GB (8GB recommended).
  • Disk: ~2GB free space (Repository + 1GB Model).
  • GPU: Optional.
    • macOS: Supports Metal acceleration (though configured to CPU-fallback by default for stability).
    • NVIDIA: Supports CUDA if configured.
  • CPU: Any modern CPU (AVX2 support recommended).

git clone https://github.com/YourUsername/ResumeParser.git
cd ResumeParser
python -m venv env
source env/bin/activate  # Windows: env\Scripts\activate
pip install -r requirements.txt

We use a script to automatically verify and download the GGUF model from Hugging Face.

python download_model.py qwen

This will download qwen2.5-1.5b-instruct-q4_k_m.gguf to resume_parser/models/.

python resume_parser/manage.py makemigrations
python resume_parser/manage.py migrate
python resume_parser/manage.py runserver

Visit http://127.0.0.1:8000/ to access the GUI.

docker-compose up --build

Note: The Docker image includes the model download step automatically.

You can create a standalone executable that doesn't require Python to be installed on the target machine.

./build_mac.sh

The output zip ResumeParser_Mac.zip will be generated in the project root.

The AI extracts data in the following JSON structure:

{
    "name": "Omkar Pathak",
    "email": "omkarpathak27@gmail.com",
    "mobile_number": "xxxxxxx34",
    "skills": ["Python", "Django", "Machine Learning", "AI"],
    "education": "BE in Computer Science, Pune University",
    "experience": "5+ years in Full Stack Development...",
    "company_names": ["Ellicium Solutions", "Numerator"],
    "ai_summary": "Highly skilled Python Developer with...",
    "ai_strengths": "Strong problem-solving, System Design..."
}

MIT License