Scorj

A tool that scores resumes against job descriptions, providing detailed feedback and analytics on resume-job fit.

👉 Click here for a quick product demo.

Features

• Intelligent resume parsing from PDF and DOCX formats
• Advanced job description analysis and requirement extraction
• AI-powered resume scoring with comprehensive feedback
• Semantic skills matching using embeddings and fuzzy logic
• Modern React frontend with interactive user interface
• RESTful API backend built with FastAPI

Project Structure

scorj/
├── backend/           
├── frontend/          
├── utils/             
├── tests/             
├── config.py         
├── requirements.txt  
└── Procfile           

Installation

1. Clone the repository:

git clone https://github.com/sohv/scorj.git
cd scorj

2. Create a virtual environment:

python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

3. Install dependencies:

pip install -r requirements.txt

4. Set up environment variables:

# create .env file with:
OPENAI_API_KEY=your_openai_api_key_here

Usage

Start the FastAPI backend:

uvicorn backend.main:app --reload --host 0.0.0.0 --port 8000

The API will be available at http://localhost:8000

Start the React frontend:

cd frontend
npm install
npm start

The web interface will be available at http://localhost:3000

Testing

Run the test suite:

python run_tests.py all

Available test commands:

• unit - Run unit tests only
• integration - Run integration tests
• all - Run all tests
• coverage - Run tests with coverage report

Core Components

• Resume Parser: Extracts structured data from PDF/DOCX files
• Job Parser: Analyzes job descriptions and extracts requirements
• Skills Matcher: Uses embeddings and fuzzy matching for semantic skill analysis
• Scoring Engine: OpenAI-powered resume evaluation
• Base Scoring Engine: Structured analysis and scoring logic

Workflow

1. Resume Upload: User uploads resume in PDF or DOCX format
2. Resume Parsing: Extract structured data including:
   • Contact information
   • Work experience with dates and descriptions
   • Education details
   • Skills and technologies
3. Job Description Input: User provides job description text or URL
4. Job Analysis: Parse job requirements including:
   • Required skills and technologies
   • Experience level expectations
   • Company and role information
5. Skills Matching: Compare resume skills against job requirements using semantic matching
6. Scoring Analysis: Generate comprehensive score using OpenAI GPT model
7. Results Display: Present detailed feedback with:
   • Overall score and dimension-specific scores
   • Matching and missing skills
   • Strengths and areas for improvement
   • Actionable recommendations

Scoring Methodology

The scoring system uses a hybrid approach combining structured analysis with AI evaluation:

Structured Analysis (Base Scoring Engine)

• Skills Matching: Semantic embedding comparison with fuzzy string matching
• Experience Evaluation: Years of experience vs job requirements
• Education Assessment: Degree level scoring (Bachelor's: 60, Master's: 80, PhD: 100)
• Relevance Calculation: Domain-specific experience weighting

AI-Powered Evaluation (OpenAI Scoring Engine)

• Contextual Analysis: GPT-4o-mini evaluates resume content against job description
• Qualitative Assessment: Analyzes writing quality, achievements, and cultural fit
• Comprehensive Scoring: Generates scores across four dimensions:
  • Skills Score (40% weight)
  • Experience Score (25% weight)
  • Education Score (20% weight)
  • Domain Score (15% weight)

Score Calculation

• Final score combines structured analysis with AI evaluation
• Results include confidence level and detailed breakdown
• Recommendations provided for skill gaps and improvements
• Transparency metrics show processing time and token usage

What Makes Scorj Different

Unlike traditional resume reviewers that rely on keyword matching or basic templates, Scorj offers several unique advantages:

Advanced AI Analysis

• Contextual Understanding: Uses OpenAI GPT-4o-mini for deep content analysis beyond simple keyword matching
• Job-Specific Evaluation: Scores resumes specifically against provided job descriptions rather than generic criteria
• Qualitative Assessment: Evaluates writing quality, achievement presentation, and cultural fit indicators

Semantic Matching Technology

• Embedding-Based Matching: Uses sentence transformers to find semantically similar skills
• Fuzzy String Matching: Catches variations in skill names and terminology
• Contextual Relevance: Understands the importance of skills in the job context

Technical Architecture

• API-First Design: REST API backend enables integration with other tools
• Modern Stack: FastAPI backend with React frontend for performance and usability
• Comprehensive Testing: Extensive test cases ensure reliability and accuracy

User-Centric Features

• Actionable Feedback: Specific recommendations for improvement rather than just scores
• Skills Gap Analysis: Identifies missing skills with suggestions for development
• Processing Transparency: Shows analysis time and token usage for trust and debugging
• Multiple Formats: Supports both PDF and DOCX resume formats
• Interactive Chat: AI-powered chat assistant for resume and job-specific questions