🤖 ResumeForge-AI

AI-Powered ATS-Friendly Resume Generator for 2025 FAANG Applications

ResumeForge-AI is an intelligent resume generation system that transforms your portfolio data into perfectly optimized, ATS-friendly resumes using advanced AI techniques. Built specifically for data engineers and tech professionals targeting FAANG and top-tier companies.

✨ Features

🎯 AI-Powered Optimization

Smart Verb Variation: Automatically tracks and varies power verbs across resume (prevents "architected" 16x!)
Keyword Optimization: Generates role-specific keywords (AI/ML, Cloud, Streaming, etc.)
Metrics Coverage: Ensures 80-90% of bullets have quantified metrics
ATS Compliance: 20-point ATS scoring system with validation

📊 Intelligent Analysis

Comprehensive Scoring: 100-point grading system (Content 50 + Format 20 + Keywords 20 + Technical 10)
Power Verb Tracking: Monitors usage across 5 categories (Leadership, Technical, Optimization, Scale, Delivery)
Metrics Detection: 15+ metric patterns (%, +, K/M/B, x, TB/GB, users, etc.)
Repetition Analysis: Identifies and fixes overused words automatically

🎨 Multiple Templates

Jake Template: Clean, ATS-optimized, single-column (recommended)
Classic Template: Traditional, professional format
Easily extensible for custom templates

🚀 Smart Features

Latest 3 Experiences Only: Optimizes recent roles, skips older ones
Vector Embeddings: Uses mxbai-embed-large for semantic understanding
Interactive CLI: User-friendly prompts for job role and template selection
LaTeX → PDF: Automatic conversion with validation

📋 Requirements

System Requirements

Python 3.8+
Ollama (local LLM runtime)
LaTeX distribution (for PDF generation):
- Ubuntu/Debian: sudo apt-get install texlive-full
- macOS: brew install --cask mactex
- Windows: MiKTeX

Python Dependencies

pip install -r requirements.txt

Key dependencies:

langchain - LLM orchestration
ollama - Local LLM runtime
chromadb - Vector database
pyyaml - Data loading
pdflatex - PDF conversion

🚀 Quick Start

1. Install Ollama and Pull Model

# Install Ollama
curl -fsSL https://ollama.ai/install.sh | sh

# Pull recommended model
ollama pull qwen2.5-coder:7b-instruct-q4_K_M

# Pull embedding model
ollama pull mxbai-embed-large

2. Clone Repository

git clone https://github.com/yourusername/ResumeForge-AI.git
cd ResumeForge-AI

3. Install Dependencies

pip install -r requirements.txt

4. Configure Your Data

Edit data/portfolio_data.yml with your information:

personal:
  name: "Your Name"
  email: "your.email@example.com"
  phone: "+1-234-567-8900"
  location: "San Francisco, CA"
  linkedin: "linkedin.com/in/yourprofile"
  github: "github.com/yourusername"

summary:
  title: "Senior Data Engineer"
  years_of_experience: 5
  specializations:
    - "Real-Time Data Engineering"
    - "Cloud Architecture (AWS)"
  key_achievement: "Reduced pipeline latency by 99%"

experience:
  - company: "Tech Company"
    position: "Senior Data Engineer"
    location: "Remote"
    start_date: "2024-01-01"
    end_date: "Present"
    achievements:
      - "Built real-time pipeline processing 50M+ events daily"
      - "Reduced costs by 30% through optimization"
# ... more experiences

projects:
  - name: "Real-Time Analytics Platform"
    description: "Built streaming pipeline with Kafka and Spark"
    technologies: ["Apache Kafka", "Spark", "AWS"]
    is_featured: true
# ... more projects

5. Generate Resume

Interactive Mode (Recommended):

python main.py

Follow the prompts:

📋 SELECT TARGET JOB ROLE
1. Data Engineer
2. Software Engineer
3. Machine Learning Engineer
...

🎨 SELECT RESUME TEMPLATE
1. JAKE (Clean, ATS-optimized)
2. CLASSIC (Traditional format)

Command-Line Mode:

# Quick generation
python main.py --role "Data Engineer" --template jake

# With custom model
python main.py --role "ML Engineer" --template classic --model llama3:8b

# Skip confirmation
python main.py --role "Data Engineer" --template jake --yes

📖 Usage Examples

Example 1: Data Engineer Resume

python main.py --role "Data Engineer" --template jake --output output/data_engineer.tex

Output:

📊 OVERALL SCORE: 89.5/100 (Grade: A)
📈 Score Breakdown:
   Content Quality:      46.5/50  ✅
   Format & ATS:         20.0/20  ✅
   Keyword Optimization: 20.0/20  ✅
   Technical Depth:      10.0/10  ✅

🎯 Key Metrics:
   Bullets with Metrics: 27/29 (93.1%)
   Total Metrics: 35
   Power Verbs: 14 used
   Metrics Coverage: 17/18 (94.4%)
   Unique Verbs Used: 16

Example 2: ML Engineer Resume

python main.py --role "Machine Learning Engineer" --template classic

Example 3: List Available Templates

python main.py --list-templates

🏗️ Architecture

ResumeForge-AI/
├── main.py                      # CLI entry point
├── data/
│   └── portfolio_data.yml       # Your portfolio data
├── src/
│   ├── ai_integration.py        # AI/LLM integration (Ollama)
│   ├── resume_generator.py      # Main generator class
│   ├── content_optimizer.py     # Intelligent optimizer (verb tracking, metrics)
│   ├── keyword_generator.py     # Role-specific keyword generation
│   ├── resume_templates.py      # Template management (Jake, Classic)
│   └── resume_analyzer.py       # 100-point scoring system
├── utils/
│   ├── portfolio_data.py        # Data loading
│   ├── data_transformer.py      # Portfolio → Resume transformation
│   └── latex_to_pdf.py          # LaTeX → PDF conversion
└── output/
    ├── resume.tex               # Generated LaTeX
    └── resume.pdf               # Final PDF

🎨 Templates

Jake Template (Recommended)

Style: Clean, modern, single-column
ATS Score: 20/20
Best For: FAANG, startups, tech companies
Features: Clear sections, optimal whitespace, ATS-friendly

Classic Template

Style: Traditional, professional
ATS Score: 18/20
Best For: Corporate, enterprise roles
Features: Conservative design, detailed sections

Adding Custom Templates

Create template class in src/resume_templates.py
Inherit from BaseResumeTemplate
Implement generate_latex() method
Register in TemplateManager

class MyTemplate(BaseResumeTemplate):
    def generate_latex(self, data: ResumeData, **kwargs) -> str:
        # Your LaTeX generation logic
        return latex_content

🧠 How It Works

Phase 1: Keyword Generation

1. Analyze target job role
2. Generate tier-1 keywords (AI, ML, LLM, RAG)
3. Generate tier-2 keywords (Real-time, Kafka, Streaming)
4. Create power verb categories

Phase 2: AI Optimization

1. Track ALL verbs used (global counter)
2. For each bullet:
   - Extract current verb
   - Check if overused (3+ times)
   - Suggest fresh alternatives
   - Check metrics coverage
   - Add metrics if < 85% coverage
3. Optimize only latest 3 experiences
4. Ensure 80-90% bullets have metrics

Phase 3: Scoring

Content Quality (50 pts):
  - Quantification (15 pts): Bullets with metrics
  - Power Verbs (10 pts): Diversity across 5 categories
  - Section Balance (10 pts): Experience vs. other sections
  - Professional Language (10 pts): No pronouns, passive voice
  - No Fluff (5 pts): Avoid buzzwords

Format & ATS (20 pts):
  - No ATS killers (images, graphics, multi-column)
  - All required sections present
  - Clean LaTeX structure

Keywords (20 pts):
  - AI/ML keywords (10 pts)
  - Domain-specific keywords (10 pts)

Technical Depth (10 pts):
  - Specific technologies mentioned
  - Architecture keywords
  - Scale indicators (50M+, 10TB+)

🎯 Optimization Features

Verb Variation System

Problem: Overused verbs (e.g., "Architected" 16 times)

Solution:

Tracks verb usage globally using Counter()
Marks verbs used 3+ times as "overused"
Provides fresh alternatives from 30+ power verbs
Categorizes by type (Leadership, Technical, Optimization, Scale, Delivery)

Example:

Before: Architected, Architected, Architected...
After: Spearheaded, Engineered, Optimized, Delivered, Scaled...

Metrics Coverage

Target: 80-90% bullets with quantified metrics

Tracked Patterns:

Percentages: 99%, 99.9%
Scale: 50M+, 10TB, 500+ users
Improvements: 60% faster, 3x throughput
Reductions: from 3 hours to 1 minute

Example:

Before: Built data pipeline for processing
After: Engineered real-time pipeline processing 50M+ events with 99.9% uptime

Smart Experience Filtering

Only optimizes latest 3 experiences:

Skips older roles (4+ years ago)
Saves AI tokens
Focuses on relevant experience
Prints: "⏭️ Skipped 2 older experiences"

📊 Sample Output

================================================================================
🤖 AI-POWERED RESUME GENERATOR (2025 EDITION)
================================================================================

✅ Template: JAKE
✅ Optimizer: Enabled

🎯 TARGET ROLE: Data Engineer
🏢 TARGET: FAANG

================================================================================
PHASE 1: KEYWORD GENERATION
================================================================================
✅ Keyword Profile Generated:
   Primary Keywords: 10 (AI, ML, LLM, GenAI, RAG...)
   Secondary Keywords: 11
   Top Keywords: AI, ML, LLM, GenAI, RAG

================================================================================
PHASE 2: AI-POWERED OPTIMIZATION
================================================================================
   📝 Professional Summary...
      ✓ +2 keywords, Enhanced

   💼 Work Experience (Latest 3 only)...
      Company 1/3: DTDC Express Limited
      ✓ 5/6 bullets (+9 kw, 4 verbs, 1 metrics)

      Company 2/3: Quantiphi Analytics Solutions
      ✓ 6/6 bullets (+0 kw, 4 verbs, 0 metrics)

      Company 3/3: Cognizant Technology Solutions
      ✓ 6/6 bullets (+1 kw, 6 verbs, 5 metrics)

      ⏭️  Skipped 2 older experiences

   ============================================================================
   ✅ OPTIMIZATION SUMMARY
   ============================================================================
   Bullets Optimized:      17
   Keywords Added:         15
   Verbs Varied:           14
   Metrics Added:          6
   Metrics Coverage:       17/18 (94.4%)
   Unique Verbs Used:      16
   Top Verbs:              reduced(4), built(4), architected(3)
   ============================================================================

================================================================================
PHASE 4: COMPREHENSIVE QUALITY ANALYSIS
================================================================================
📊 OVERALL SCORE: 89.5/100 (Grade: A)

📈 Score Breakdown:
   Content Quality:      46.5/50  ✅
   Format & ATS:         20.0/20  ✅
   Keyword Optimization: 20.0/20  ✅
   Technical Depth:      10.0/10  ✅

💡 EXCELLENT RESUME:
   ✅ Strong verb diversity (16 unique)
   ✅ Comprehensive quantification (94.4%)
   ✅ ATS-optimized format
   ✅ Keyword-rich technical depth

🔧 Configuration

Model Configuration

Edit src/ai_integration.py:

DEFAULT_MODEL = "qwen2.5-coder:7b-instruct-q4_K_M"
EMBEDDING_MODEL = "mxbai-embed-large"
TEMPERATURE = 0.2  # Lower = more consistent

Optimization Settings

Edit src/content_optimizer.py:

METRICS_TARGET = 0.85  # 85% bullets with metrics
MAX_EXPERIENCES = 3    # Only optimize latest 3
VERB_OVERUSE_THRESHOLD = 3  # Flag verbs used 3+ times

Keywords Customization

Edit src/keyword_generator.py:

CRITICAL_KEYWORDS_2025 = {
    'tier1': {
        'keywords': ['AI', 'ML', 'LLM', 'GenAI', 'RAG'],
        'weight': 3.0,
        'target': (2, 3)  # Appear 2-3 times
    }
}

🐛 Troubleshooting

Ollama Connection Error

# Start Ollama service
ollama serve

# Check if model is available
ollama list

LaTeX Compilation Failed

# Install LaTeX
sudo apt-get install texlive-full

# Or use manual conversion
pdflatex output/resume.tex

Low Score (< 70/100)

Common issues:

Missing metrics → Add quantified achievements (%, +, scale)
Overused verbs → System will auto-fix on next run
Missing keywords → Add relevant tech stack to portfolio_data.yml
Weak bullets → Add specific numbers, scale, impact

📝 TODO / Roadmap

🤝 Contributing

Contributions are welcome! Please follow these steps:

Fork the repository
Create feature branch (git checkout -b feature/AmazingFeature)
Commit changes (git commit -m 'Add AmazingFeature')
Push to branch (git push origin feature/AmazingFeature)
Open Pull Request

Areas for contribution:

New resume templates
Additional keyword profiles (SWE, DevOps, etc.)
Better scoring algorithms
Additional LLM model support
Documentation improvements

📄 License

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

🙏 Acknowledgments

Ollama - Local LLM runtime
LangChain - LLM orchestration framework
Jake's Resume Template - Original LaTeX template inspiration
ChromaDB - Vector database

📧 Contact

Krishnanand Anil

LinkedIn: linkedin.com/in/krishnanand-anil
GitHub: github.com/sudo-krish
Email: krishnanandpanil@gmail.com
portfolio: krishnanandanil.com

⭐ Star History

If this project helped you land a job, please give it a star! ⭐

Made with ❤️ by Krishnanand Anil

Name		Name	Last commit message	Last commit date
Latest commit History 5 Commits
data		data
src		src
template		template
utils		utils
.gitignore		.gitignore
README.md		README.md
main.py		main.py
requirements.txt		requirements.txt

sudo-krish/ResumeForge-AI

Folders and files

Latest commit

History

Repository files navigation