VACAI - AI-Powered Job Search Automation

Automate your job search with AI-powered job matching and scoring. VACAI analyzes your resume, scrapes job listings from multiple platforms, and scores each position based on your skills and preferences.

Features

• 🤖 AI Resume Analysis: Automatically extract your skills and preferences using OpenAI GPT-4
• 📄 PDF Support: Direct PDF resume parsing (also supports .txt and .md)
• 🔍 Multi-Platform Scraping: Search across LinkedIn, Indeed, Glassdoor, and more
• 📊 Smart Scoring: Single AI agent scores jobs holistically (token-efficient)
• 📈 Daily Reports: Get ranked job matches delivered automatically
• 📱 Telegram Integration: Receive daily reports with clickable "Apply" links directly in Telegram

Quick Start

Option 1: Docker (Recommended)

# Pull the latest image (replace Neverdecel with the repo owner)
docker pull ghcr.io/Neverdecel/vacai:latest

# Create directories and .env file
mkdir -p data reports logs config resume
cp config/.env.example .env
# Edit .env and add your OPENAI_API_KEY

# Initialize with your resume
docker run --rm \
  -v $(pwd)/data:/app/data \
  -v $(pwd)/config:/app/config \
  -v $(pwd)/resume:/app/resume \
  --env-file .env \
  ghcr.io/Neverdecel/vacai:latest \
  python main.py init --resume /app/resume/your_resume.pdf

# Run a scan
docker run --rm \
  -v $(pwd)/data:/app/data \
  -v $(pwd)/config:/app/config \
  -v $(pwd)/reports:/app/reports \
  --env-file .env \
  ghcr.io/Neverdecel/vacai:latest \
  python main.py scan

# Or use docker-compose
docker-compose up -d
docker-compose exec vacai python main.py scan

See docs/DEPLOYMENT.md for full Docker deployment guide.

Option 2: Local Installation

1. Install dependencies

pip install -r requirements.txt

2. Setup configuration

cp config/.env.example .env
# Add your OPENAI_API_KEY

3. Initialize with your resume

python main.py init --resume path/to/resume.pdf

4. Run job scan

python main.py scan

5. View results

python main.py report

6. Setup Telegram notifications (optional)

# See TELEGRAM_SETUP.md for detailed instructions
python main.py test-telegram

How It Works

1. Analyze Resume: AI extracts your skills and preferences from your resume
2. Scrape Jobs: Automatically searches LinkedIn, Indeed, Glassdoor, etc.
3. Score & Rank: AI evaluates each job against your profile
4. Daily Reports: Get ranked results with scores and reasoning
5. Notifications: Optional Telegram integration for daily updates

Architecture

• Single AI Agent: One OpenAI API call per job for efficient token usage
• JobSpy Integration: Scrapes jobs from multiple platforms
• SQLite Database: Stores jobs and scores locally
• Structured Output: Pydantic models ensure consistent scoring
• PDF Support: PyPDF2 for resume text extraction

Cost Estimate

• ~100 jobs × 500 tokens = 50K tokens per scan
• Using GPT-4o-mini: ~$0.01-0.02 per scan
• Using GPT-4o: ~$0.25 per scan
• Resume analysis (one-time): ~$0.01-0.05

Contributing

We welcome contributions! Please see CONTRIBUTING.md for guidelines.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Support

• Issues: Report bugs or request features via GitHub Issues
• Documentation: See /docs folder for detailed guides
• Security: See SECURITY.md for reporting vulnerabilities

Deployment

Docker & GHCR

VacAI is available as a Docker image on GitHub Container Registry:

docker pull ghcr.io/neverdecel/vacai:latest

Available tags:

• latest - Latest stable release
• main - Main branch builds
• v1.0.0 - Specific version tags

Quick deploy with docker-compose:

docker-compose up -d

See docs/DEPLOYMENT.md for comprehensive deployment guide including:

• Docker setup and configuration
• GHCR image usage
• Scheduled scans with cron
• Production best practices
• Health checks and monitoring

CI/CD

Automated builds on:

• Push to main branch
• Version tag releases
• Pull request validation

Images are automatically built for linux/amd64 and linux/arm64 platforms.