I Built an AI That Writes My Job Applications - And It's Open Source

Source: Dev.to

The Problem

Job applications are painful. Most engineers go through these steps for every role:

1. Read the JD – Extract requirements, keywords, and company culture.
2. Tailor resume – Rewrite bullets to match the specific role.
3. Write cover letter – Explain why you’re perfect for the position.
4. Research company – Understand values, recent news, etc.
5. Prep for interview – Anticipate questions and prepare answers.
6. Follow up – Send thank‑you emails and handle rejections.

Each application can take 2–4 hours. Many people skip steps 2‑5 and send generic materials, which drives up rejection rates.

The Solution: Career Architect

Career Architect is an AI‑driven pipeline that automates the entire application process. It works with any LLM (Claude, GPT‑4, Cursor, etc.).

# Clone and install
git clone https://github.com/henryohanga/career-architect
cd career-architect
make install

# Paste a job description to the AI assistant
# AI generates everything for you
python scripts/compile_all.py   # Build PDFs

How It Works

Step 1 – Build Your Experience Lake

Create a single source of truth for your professional background:

source_materials/
├── identity.json          # Contact info, preferences
├── master_experience.md   # All your achievements
├── resumes/               # Historical resume versions
│   ├── 2024-current.md
│   └── 2023-previous.md
└── projects/              # Key project details
    ├── saas-platform.md
    └── api-redesign.md

Prompt to the AI:

“Read the setup prompt and analyze my resumes to build master_experience.md.”

The assistant extracts achievements, adds metrics, and structures everything using the Modern Builder Framework.

Step 2 – Apply to Jobs

Paste a job description and tell the AI what you need:

“I want to apply for this Senior Engineer role. Use the Career Architect pipeline.”

The AI creates a new folder, e.g., applications/2025-01-07-company-role/, and:

• Analyzes gaps between your experience and the role’s requirements.
• Generates a tailored resume, cover letter, and interview prep material.
• Saves everything as Markdown files.

Step 3 – Build PDFs

One command converts the Markdown output into polished PDFs:

python scripts/compile_all.py
# Output files: resume.pdf, cover_letter.pdf

The Tech Stack

Key Scripts

• build_resume.py – Markdown → formatted PDF.
• compile_all.py – Batch processing with progress tracking.
• career.py – Unified CLI interface.

The AI Magic: Prompt Engineering

The real innovation lies in the prompt system. Below is an excerpt of the resume‑tailoring prompt used by the assistant:

# Role: Career Branding Expert
...

(Full prompt omitted for brevity; see the repository for the complete prompt library.)

Inputs

• source_materials/master_experience.md
• applications/[folder]/job_desc.md
• source_materials/identity.json (for preferences)

Configuration

Source: identity.json → preferences

Note: Apply the appropriate style guide from style_guide.md.

Resume Styles

Modern Builder (Tech Startups)

Modern Builder Capabilities
- Systems Thinking: Architected event‑driven architecture processing **50 K events/sec**
- Technical Taste: Implemented zero‑downtime deployments reducing MTTR by **75 %**
- Ownership: Led incident response restoring service in **12 minutes**

Traditional (Enterprise)

Professional Summary
Experienced software engineer with **6+ years** building scalable systems...

Work Experience
Senior Software Engineer | TechCorp | 2020‑Present
- Led development of React SPA serving **50 000 daily users**
- Managed team of **5 engineers** improving sprint velocity by **35 %**

Multi‑Language Support

Set your language preference in identity.json:

{
  "preferences": {
    "language": "de",
    "resume_style": "traditional"
  }
}

The system automatically translates section headers and adapts tone:

Extending the System

Adding New Resume Styles

Create a new style in .prompts/style_guide.md:

### Style: `minimalist`

**Best for:** Design‑focused roles, creative positions  

**Language Patterns**
1. Clean, concise bullets  
2. Focus on impact over process  
3. Visual/portfolio emphasis  

Custom Prompts

Add industry‑specific prompts under .prompts/:

.prompts/
├── cybersecurity.md   # Security‑specific language
├── fintech.md          # Finance terminology
└── gaming.md           # Game‑dev focus

Integration Examples

GitHub Actions Workflow

name: Build Job Application
on:
  push:
    paths:
      - 'applications/**'

jobs:
  build:
    runs-on: ubuntu-latest
    steps:
      - uses: actions/checkout@v3
      - name: Build PDFs
        run: python scripts/compile_all.py
      - name: Upload artifacts
        uses: actions/upload-artifact@v3
        with:
          name: application-materials
          path: applications/**/*.pdf

Real Results

User‑reported improvements

Getting Started

Prerequisites

1. Python 3.8+
2. An AI assistant (Claude, GPT‑4, Cursor, etc.)
3. Pandoc + LaTeX (for PDF generation)

Quick Start

# Clone and set up the repo
git clone https://github.com/henryohanga/career-architect
cd career-architect
make install

# Configure your profile
#   • Edit source_materials/identity.json
#   • Add historical resumes to source_materials/resumes/

# Build your experience lake
# Prompt the AI: “Use Career Architect setup prompt”

# Generate applications
# Paste a JD to the AI: “Use Career Architect for this role”
python scripts/compile_all.py

The Philosophy

1. AI as a Collaborator – AI assistants act as senior partners; prompts are crafted to exploit AI strengths while keeping human judgment central.
2. Modern Builder Framework – Emphasizes ownership, systems thinking, and measurable impact—key traits for senior engineering roles.
3. Unopinionated by Design – Supports multiple styles and languages; there is no single “right way” to write a resume.

Challenges & Learnings

1. Prompt Engineering is Hard – Effective prompts require deep knowledge of AI limits and human psychology; what works for GPT‑4 may fail with Claude.
2. Balancing Automation vs. Personalization – Over‑automation yields generic output; the goal is structured guidance that still lets personality shine.
3. Multi‑Language Complexity – Translation is only part of the story; cultural adaptation of professional norms is essential.

Future Roadmap

1. Web Interface – Friendly UI for non‑technical users.
2. ATS Optimization – Advanced keyword matching for applicant‑tracking systems.
3. A/B Testing – Compare different resume versions automatically.
4. Team Features – Managers can assist team members in building their applications.
5. Integration APIs – Connect with LinkedIn, Indeed, and other job platforms.

Call to Action

If you’re tired of spending hours on job applications, give Career Architect a try:

1. Star the repo – Show your support.
2. Try it out – Run it against a sample job description.
3. Contribute – Add new styles, languages, or features.
4. Share – Help other developers land better jobs.

• GitHub:
• Product Hunt: Coming Soon

Built with ❤️ by a developer, for developers. MIT licensed.

SEO Keywords

• AI job applications
• Resume automation
• AI prompt engineering
• Career development tools
• Open source developer tools
• Job search automation