AI-powered resume tailoring system that adapts your base resume to specific job descriptions while preserving truthfulness.

This system intelligently tailors resumes to job postings by:

• Analyzing job requirements and your resume
• Scoring relevance of each bullet point
• Rewriting content for clarity (without fabrication)
• Filtering low-relevance bullets
• Reordering sections by relevance
• Generating tailored resumes in multiple formats

Key Principle: Preserve truth, improve signal-to-noise ratio.

• PDF Parsing: Layout-aware extraction preserving structure
• LLM Integration: Gemini and DeepSeek support with structured outputs
• Job Detection: Auto-extracts requirements from job descriptions
• Alignment Scoring: 0-1 relevance score per bullet with explanations
• Controlled Rewriting: Strict no-fabrication rules with hallucination detection
• Quality Validation: Automated checks for metric consistency, keyword stuffing
• Multiple Formats: Output as PDF, DOCX, or Markdown
• Chrome Extension: One-click tailoring from job boards (LinkedIn, Indeed, Greenhouse, Lever)
• REST API: Full-featured API with auto-generated docs
• Docker Support: Easy deployment locally or to cloud

• Python 3.11+
• uv (Python package manager)
• API key for Gemini or DeepSeek
• Docker (optional, for containerized deployment)

1. Clone and setup:

cd backend
uv venv
source .venv/bin/activate  # On macOS/Linux
# On Windows: .venv\Scripts\activate
uv pip install -r requirements.txt

2. Configure API keys:

cp .env.example .env
# Edit .env and add your API keys

3. Test the system:

# Parse a resume
python parse_resume.py path/to/resume.pdf

# Extract job requirements
python extract_job.py sample_job.txt

# Score alignment
python score_alignment.py path/to/resume.pdf sample_job.txt

# Complete pipeline
python assemble_resume.py path/to/resume.pdf sample_job.txt

cp .env.example .env
# Edit .env with API keys
./start-docker.sh

Access API at http://localhost:8000/docs

Parse Resume:

python parse_resume.py resume.pdf

Extract Job Requirements:

python extract_job.py job_description.txt
python extract_job.py --text "Job description text here"

Score Alignment:

python score_alignment.py resume.pdf job_description.txt

Rewrite Resume:

python rewrite_resume.py resume.pdf job_description.txt
python rewrite_resume.py resume.pdf job_description.txt gemini 0.6

Complete Pipeline:

python assemble_resume.py resume.pdf job_description.txt
python assemble_resume.py resume.pdf job_description.txt --format pdf --threshold 0.4

Quality Validation:

python validate_quality.py resume.pdf job_description.txt

Start server:

cd backend
./start_server.sh

API Endpoints:

• POST /api/parse-resume - Parse PDF resume
• POST /api/extract-job - Extract job requirements
• POST /api/score-alignment - Score resume-job alignment
• POST /api/tailor-resume - Complete tailoring pipeline

Documentation:

• Swagger UI: http://localhost:8000/docs
• ReDoc: http://localhost:8000/redoc

See backend/API_EXAMPLES.md for detailed usage.

1. Install extension:

   • Open chrome://extensions/
   • Enable "Developer mode"
   • Click "Load unpacked"
   • Select extension folder

2. Start backend:

   cd backend
   ./start_server.sh

3. Use extension:

   • Navigate to job posting (LinkedIn, Indeed, etc.)
   • Click extension icon
   • Upload resume PDF
   • Click "Tailor Resume"
   • Download tailored resume

See extension/README.md for detailed instructions.

┌─────────────────────────────────────────────────────────────┐
│                    Chrome Extension (MV3)                    │
│  ┌──────────────────┐         ┌──────────────────────────┐  │
│  │  Content Script  │────────▶│   Popup UI / Service     │  │
│  │  (Job Detection) │         │   Worker (API Client)    │  │
│  └──────────────────┘         └──────────────────────────┘  │
└────────────────────────────────────┬────────────────────────┘
                                     │ REST API
                                     ▼
┌─────────────────────────────────────────────────────────────┐
│                   FastAPI Backend (Python)                   │
│  ┌──────────────────────────────────────────────────────┐   │
│  │              Resume Processing Pipeline               │   │
│  │  1. Parse     →  2. Extract  →  3. Score             │   │
│  │       ↓              ↓              ↓                 │   │
│  │  4. Rewrite   →  5. Assemble →  6. Validate          │   │
│  └──────────────────────────────────────────────────────┘   │
│  ┌──────────────────┐  ┌──────────────────────────────┐    │
│  │  PDF Parser      │  │  LLM Orchestrator            │    │
│  │  (PyMuPDF/       │  │  (Gemini/DeepSeek +          │    │
│  │   pdfplumber)    │  │   Instructor)                │    │
│  └──────────────────┘  └──────────────────────────────┘    │
└─────────────────────────────────────────────────────────────┘

backend/
├── app/
│   ├── models/          # Pydantic data models
│   ├── parsers/         # PDF and text parsing
│   ├── llm/            # LLM integration
│   ├── pipeline/       # Processing pipeline
│   └── api/            # FastAPI endpoints
├── tests/              # Test suite
├── *.py               # CLI tools
└── requirements.txt    # Python dependencies

extension/
├── manifest.json       # Extension configuration
├── content.js         # Job detection
├── background.js      # Service worker
├── popup.html/js      # UI and logic
└── config.js          # Environment config

cd backend
pytest tests/

See extension/TESTING.md for comprehensive testing checklist.

• Unit tests for parsers, models, and pipeline components
• Integration tests for API endpoints
• Manual testing guide for extension

./start-docker.sh

See DEPLOYMENT.md for detailed guides:

• AWS ECS + Fargate ($10-30/month)
• GCP Cloud Run ($0-10/month with free tier)
• Azure Container Instances ($10-20/month)

Extension supports multiple environments:

• Local: http://localhost:8000
• Staging: Configure in extension settings
• Production: Configure in extension settings

• API keys stored in environment variables
• Secrets management for cloud deployment
• CORS configured for extension
• Input validation on all endpoints
• Quality validation prevents bad outputs

LLM Costs (per resume):

• Gemini Flash: ~$0.01-0.02
• DeepSeek: ~$0.005-0.01

Infrastructure:

• Local: Free (your machine)
• Cloud: $0-30/month depending on platform and usage

Tips:

• Use DeepSeek for lower costs
• Cloud Run scales to zero (pay per use)
• Set relevance threshold higher to reduce processing

"No job detected"

• Refresh page and wait 2-3 seconds
• Use manual input fallback

"Failed to tailor resume"

• Check backend is running
• Verify API keys are configured
• Check quality validation didn't fail

"Quality validation failed"

• Lower relevance threshold
• Check for metric changes in rewrite
• Review validation report

Extension can't connect

• Verify backend URL in extension
• Check CORS configuration
• Look at browser console for errors

See DEPLOYMENT.md for more troubleshooting tips.

• API Examples: backend/API_EXAMPLES.md
• Extension Guide: extension/README.md
• Extension Testing: extension/TESTING.md
• Deployment Guide: DEPLOYMENT.md

1. Parse Resume: Extract structured data from PDF
2. Extract Job: Analyze job description for requirements
3. Score Alignment: Rate each bullet's relevance (0-1)
4. Rewrite Bullets: Improve clarity without fabrication
5. Assemble Resume: Filter, reorder, and format
6. Validate Quality: Check for hallucinations and issues

• No Fabrication: Strict rules prevent inventing experience
• Metric Consistency: Detects if numbers were added/changed
• Hallucination Detection: Multiple checks for false content
• Content Loss Prevention: Warns if too many bullets removed
• Keyword Stuffing Detection: Flags unnatural repetition

This is a personal project, but suggestions are welcome!

MIT License - See LICENSE file for details

Built with:

• FastAPI for the API
• Instructor for structured LLM outputs
• PyMuPDF & pdfplumber for PDF parsing
• ReportLab for PDF generation
• Google Gemini & DeepSeek for LLM capabilities

For issues or questions:

1. Check documentation in respective folders
2. Review troubleshooting section
3. Check browser/server console logs

Status: Production Ready ✅

All 12 planned tasks completed. System is fully functional for local use and ready for cloud deployment.