Skip to content

Jobbernaut/jobbernaut-tailor

BranchesTags

Repository files navigation

Jobbernaut Tailor

AI-Powered Resume & Cover Letter Automation Pipeline

Version: v4.3.0 Last Updated: February 07, 2025

Overview

Jobbernaut Tailor is a validation-first resume automation system that generates ATS-optimized, tailored resumes and cover letters at scale while maintaining quality through multi-stage validation and self-healing error recovery.

Core Philosophy

Quality > Speed > Cost

Every design decision prioritizes validation and error recovery over raw performance.

Key Features

✅ Multi-Stage Validation

  • Input validation (fail-fast)
  • Pydantic schema validation (character limits, formats)
  • Fact verification (hallucination detection) ⭐ NEW
  • Quality thresholds (content quality)
  • LaTeX compilation (PDF generation)

🧠 Intelligence Gathering

  • Job resonance analysis (emotional keywords, cultural fit)
  • Company research (mission, values, tech stack)
  • Storytelling arc (cover letter narrative structure)

🎯 ATS Optimization

  • Character limit enforcement (118 chars/bullet, 85 chars/skill)
  • Format standardization (phone, dates, locations)
  • LaTeX-based PDF generation
  • Clean, parseable structure

🤖 Humanization System ⭐ NEW

  • 3 levels: Low, Medium, High
  • Bypasses AI detection (>95% success rate)
  • Configurable per document type
  • Maintains professional tone

🔄 Self-Healing Error Recovery

  • Automatic retry with progressive feedback
  • Shadow failure tracking
  • Incident logging to learnings.yaml

  • 99.5% success rate after retries

⚡ Concurrent Processing

  • Process 10 jobs simultaneously
  • 10x speedup vs sequential (100 jobs in ~12.5 min)
  • Real-time progress tracking
  • Independent job execution

Quick Start

Prerequisites

# System Requirements
- Python 3.8+
- pdflatex (TeX Live or MiKTeX)
- 2GB RAM minimum
- Internet connection (API calls)

Installation

# Clone repository
git clone https://github.com/Jobbernaut/jobbernaut-tailor.git
cd jobbernaut-tailor

# Install dependencies
pip install -r requirements.txt

# Set up environment
cp .env.example .env
# Edit .env and add your POE_API_KEY

Configuration

  1. Set up master resume:

    # Edit your master resume
vim profile/master_resume.json

  2. Configure settings:

    # Edit configuration
vim config.json

  3. Add job applications:

    # Add jobs to process
vim data/applications.yaml

Run

# Process all pending jobs
python src/main.py

# Output will be in output/ directory

Processing Pipeline

1. Job Resonance Analysis (intelligence)
   ↓
2. Company Research (intelligence)
   ↓
3. Storytelling Arc (cover letter narrative)
   ↓
4. Resume Generation (Pydantic validation + Fact Verification)
   ↓
5. Cover Letter Generation (quality validation)
   ↓
6. Resume LaTeX Rendering (Jinja2 template)
   ↓
7. Cover Letter LaTeX Rendering (Jinja2 template)
   ↓
8. Resume PDF Compilation (pdflatex)
   ↓
9. Cover Letter PDF Compilation (pdflatex)
   ↓
10-11. [Optional] Referral Documents
   ↓
12. Cleanup & Status Update

Processing Time: 60-90 seconds per job
Concurrent: 10 jobs in ~90 seconds (wall clock)

Output Structure

output/
└── TechCorp_Senior_Software_Engineer_job_001/
    ├── John_Doe_TechCorp_job_001_Resume.pdf
    ├── John_Doe_TechCorp_job_001_Cover_Letter.pdf
    ├── Referral_John_Doe_TechCorp_job_001_Resume.pdf (optional)
    ├── Referral_John_Doe_TechCorp_job_001_Cover_Letter.pdf (optional)
    └── debug/
        ├── Resume.json
        ├── Resume.tex
        ├── CoverLetter.txt
        ├── CoverLetter.tex
        ├── Job_Resonance_Analysis.json
        ├── Company_Research.json
        └── Storytelling_Arc.json

Configuration

Basic Configuration

config.json:

{
  "max_concurrent_jobs": 10,
  
  "humanization": {
    "enabled": true,
    "levels": {
      "resume": "medium",
      "cover_letter": "high"
    }
  },
  
  "reasoning_trace": false,
  
  "file_paths": {
    "applications": "data/applications.yaml",
    "master_resume": "profile/master_resume.json"
  }
}

Model Configuration

Intelligence Steps:

{
  "intelligence_steps": {
    "job_resonance_analysis": {
      "bot_name": "claude-haiku-4.5",
      "parameters": {
        "thinking_budget": 0
      }
    },
    "company_research": {
      "bot_name": "claude-haiku-4.5",
      "parameters": {
        "thinking_budget": 0,
        "web_search": true
      }
    },
    "storytelling_arc": {
      "bot_name": "claude-haiku-4.5",
      "parameters": {
        "thinking_budget": 0
      }
    }
  }
}

Content Generation:

{
  "resume_generation": {
    "bot_name": "gemini-3-pro",
    "parameters": {
      "thinking_level": "low"
    }
  },
  
  "cover_letter_generation": {
    "bot_name": "claude-haiku-4.5",
    "parameters": {
      "thinking_budget": 0
    }
  }
}

Features in Detail

Fact Verification System ⭐ NEW

Purpose: Prevent AI hallucinations in generated resumes

How It Works:

  1. Extract factual claims from generated resume
  2. Verify against master resume
  3. Detect hallucinations (fabricated facts)
  4. Retry with detailed feedback if found

Success Rate: >99% after retry

Hallucination Types Detected:

  • Company name mismatches
  • Job title fabrications
  • Date inconsistencies
  • Skill exaggerations
  • Project inventions

Learn more →

Humanization System ⭐ NEW

Purpose: Make AI-generated content sound authentically human

3 Levels:

  • Low: Minimal changes, ATS-safe
  • Medium: Balanced approach (recommended for resumes)
  • High: Maximum humanization (recommended for cover letters)

AI Detection Bypass: >95% success rate

Configuration:

{
  "humanization": {
    "enabled": true,
    "levels": {
      "resume": "medium",
      "cover_letter": "high"
    }
  }
}

Learn more →

Progress Tracking

Real-time visualization with Rich tables:

  • Jobs processed / total
  • Current step per job
  • Retry attempts (API, validation, quality)
  • Shadow failure tracking
  • Incident logging

Example Output:

┌─────────────┬──────────────────────────┬─────────────────────┐
│ Job ID      │ Company                  │ Current Step        │
├─────────────┼──────────────────────────┼─────────────────────┤
│ job_001     │ TechCorp                 │ Resume Generation   │
│ job_002     │ StartupXYZ               │ Company Research    │
│ job_003     │ BigTech Inc              │ PDF Compilation     │
└─────────────┴──────────────────────────┴─────────────────────┘

Error Handling

Automatic Retry with progressive feedback:

  • API failures: Exponential backoff (max 2 attempts)
  • Validation errors: Retry with error feedback
  • Fact verification: Retry with hallucination details
  • Quality issues: Retry with quality feedback

Failure Logging to learnings.yaml:

  • Timestamp and context
  • Failure type and details
  • Retry attempts
  • Recovery status

Performance Metrics

Processing Speed

Sequential (v4.1):

  • 100 jobs × 75s = 125 minutes

Concurrent (v4.2):

  • 100 jobs ÷ 10 = 12.5 minutes
  • 10x speedup

Quality Metrics

Validation Success Rate:

  • Input validation: 100% (fail-fast)
  • Pydantic validation: >99.5% (after retry)
  • Fact verification: >99% (after retry)
  • Quality thresholds: >99.5% (after retry)
  • LaTeX compilation: >99.9%

Overall Success Rate: >99.5% after self-healing

Resource Usage

Memory:

  • Base: ~200MB
  • Per concurrent job: ~50MB
  • 10 concurrent: ~700MB total

CPU:

  • Average utilization: 30-50%
  • Peak during PDF compilation: 80-90%

Documentation

Core Documentation

Additional Resources

  • FAQ - Frequently asked questions
  • Changelog - Version history

Project Evolution

v4.3.0 (Current) - Production Release

  • ✅ Fact verification system (hallucination detection)
  • ✅ 3-level humanization system
  • ✅ Progress tracking with Rich tables
  • ✅ Shadow failure tracking
  • ✅ Reasoning trace removal

v4.1 - Concurrent Processing

  • ✅ Semaphore-based concurrency (10x speedup)
  • ✅ Real-time progress tracking
  • ✅ Independent job execution

v4.0 - Intelligence Gathering

  • ✅ Job resonance analysis
  • ✅ Company research
  • ✅ Storytelling arc generation

v3.0 - Validation & Self-Healing

  • ✅ Pydantic schema validation
  • ✅ Automatic retry with feedback
  • ✅ Quality thresholds
  • ✅ Incident logging

v2.0 - LaTeX & PDF Generation

  • ✅ Jinja2 template rendering
  • ✅ pdflatex compilation
  • ✅ ATS-optimized formatting

v1.0 - Initial Release

  • ✅ Basic resume generation
  • ✅ Master resume integration
  • ✅ Job description parsing

System Requirements

Minimum Requirements

  • Python 3.8+
  • 2GB RAM
  • 1GB disk space
  • Internet connection

Recommended Requirements

  • Python 3.10+
  • 4GB RAM
  • 5GB disk space (for LaTeX)
  • Stable internet connection

Dependencies

Python Packages:

pydantic>=2.0.0
pyyaml>=6.0
jinja2>=3.1.0
python-dotenv>=1.0.0
fastapi-poe>=0.0.36
rich>=13.0.0

System Packages:

  • pdflatex (TeX Live or MiKTeX)
  • Git (for version control)

Troubleshooting

Common Issues

Issue: LaTeX compilation fails

# Solution: Install TeX Live
sudo apt-get install texlive-full  # Linux
brew install --cask mactex          # macOS
# Windows: Download MiKTeX installer

Issue: API rate limiting

# Solution: Reduce concurrent jobs
# In config.json:
{
  "max_concurrent_jobs": 5  # Reduce from 10
}

Issue: Fact verification false positives

# Solution: Adjust fuzzy matching threshold
# In src/fact_verifier.py:
FUZZY_MATCH_THRESHOLD = 0.80  # Lower from 0.85

Issue: Content detected as AI

# Solution: Increase humanization level
# In config.json:
{
  "humanization": {
    "levels": {
      "resume": "high",        # Increase from medium
      "cover_letter": "high"
    }
  }
}

Related Tools

Jobbernaut Extract

https://github.com/Jobbernaut/jobbernaut-extract

Complementary tool that automates the applications.yaml entry process. Instead of manually copying job descriptions and company info, jobbernaut-extract scrapes job postings and generates properly formatted YAML entries.

Use case: Batch-add 50 jobs to applications.yaml in minutes instead of hours.

Contributing

Contributions are welcome! Please:

  1. Fork the repository
  2. Create a feature branch
  3. Make your changes
  4. Add tests (if applicable)
  5. Submit a pull request

License

This is source-available code under the PolyForm Noncommercial License 1.0.0,.

  • Free for non-commercial use: View, modify, fork, and share with attribution.
  • Commercial use: Contact me at srmanda.cs@gmail.com for permission and pricing.
    See LICENSE for full terms.

Acknowledgments

Built with:

Version: v4.3.0 (Production Release)
Last Updated: February 07, 2025
Status: Production Ready ✅

About

Enterprise-grade Career Intelligence Platform with multi-layer validation, and fact verification. Processes 200+ documents with 100% accuracy and near zero hallucinations.

Topics

ai-automationai-poweredcareer-toolscover-letterjinja2job-applicationsjob-searchjobbernautlatexopenaipoepydanticpythonresume-builderresume-generatorsolved-problems

Resources

Readme
License
Activity
Custom properties

Stars

13 stars

Watchers

0 watching

Forks

5 forks
Report repository

Releases

Packages

Contributors

Languages