[GitHub] MadsLorentzen/ai-job-search
MadsLorentzen/ai-job-search is an open-source, AI-powered job application framework built on Claude Code that automates the full job search workflow from scraping to interview prep The framework uses a drafter-reviewer pipeline where Claude evaluates job fit, drafts tailored CVs (LaTeX) and cover letters, and critiques outputs before finalization Built by a geophysicist who was laid off in late 2025, it produced 69 tailored applications, 20 first interviews, and one job offer as an AI engineer b
Analysis
TL;DR
- MadsLorentzen/ai-job-search is an open-source, AI-powered job application framework built on Claude Code that automates the full job search workflow from scraping to interview prep
- The framework uses a drafter-reviewer pipeline where Claude evaluates job fit, drafts tailored CVs (LaTeX) and cover letters, and critiques outputs before finalization
- Built by a geophysicist who was laid off in late 2025, it produced 69 tailored applications, 20 first interviews, and one job offer as an AI engineer by June 2026
- The core workflow is language- and country-agnostic, with Danish job portal skills (Jobindex, Jobnet, etc.) as the default that can be swapped for local boards
- Supports optional salary benchmarking, ATS parseability checks via pdftotext, and is compatible with alternative agent tools like Codex, Antigravity, and Gemini CLI through community forks
Why It Matters
This project demonstrates a practical, real-world application of agentic AI workflows beyond typical coding assistants—showing that structured AI automation can meaningfully impact career outcomes. For AI practitioners, it serves as a compelling case study in designing multi-stage agent pipelines (scrape → evaluate → draft → review → finalize) with human-in-the-loop checkpoints. The transparency of the creator sharing both methodology and results (69 applications, 20 interviews, 1 offer) provides rare empirical evidence for the effectiveness of AI-augmented professional workflows.
Technical Details
- Architecture: Multi-agent workflow using Claude Code CLI with a drafter-reviewer pattern; the reviewer agent critiques drafts before final output, encoding career guidance best practices
- Job Portal Integration: Built-in skills for Danish job boards (Jobindex, Jobnet, Akademikernes Jobbank, LinkedIn, FreeHire) using Bun/TypeScript CLI tools; the pattern is designed to be modular and swappable for other markets
- Output Generation: LaTeX-based CV compilation (lualatex) and cover letter generation (xelatex with cover.cls and fontspec); includes optional ATS parseability checking via poppler's pdftotext
- Tech Stack: Python 3.10+, Bun runtime for CLI tools, LaTeX distributions (TeX Live, MacTeX, TinyTeX, MiKTeX); AGENTS.md provides compatibility guidance for alternative agent frameworks
- Workflow Pipeline:
/setup(profile creation from documents folder or manual entry) →/scrape(job search and fit scoring) →/apply(draft CV + cover letter with reviewer critique loop)
Industry Insight
- The rise of personal AI job search frameworks signals a shift toward individual-level AI automation that bypasses traditional recruiting channels; professionals should expect and adapt to a landscape where AI-augmented applicants become the norm
- The drafter-reviewer pipeline pattern demonstrated here is directly transferable to other high-stakes document generation tasks (proposals, reports, legal documents) where quality control through iterative critique is essential
- The creator's transparency about using AI tools with employers—resulting in positive technical conversations rather than rejection—suggests that ethical disclosure of AI assistance may become a competitive advantage rather than a liability in professional settings
Disclaimer: The above content is generated by AI and is for reference only.