job-search-project

An AI-driven job search system for entry-level Data Science / ML Engineer / Software Engineer roles. It turns a job posting into a tailored, one-page, ATS-ready application packet — resume, optional cover letter, the posting's real screening questions, and draft referral outreach — and can also run itself daily to go find postings in the first place.

This project extends MadsLorentzen/ai-job-search, which established the core pattern this is built on: a Claude Code agent pipeline tailoring a LaTeX resume/cover letter against a job description and tracking applications in a CSV. I did not design that base pattern — see Credit for what's upstream versus what I added.

What this is / what it isn't

This is a verification-first, human-in-the-loop assistant. It never submits a form, never autofills an application, and never creates an account. Every run — single posting or daily batch — ends by handing back a packet for the user to review and submit themselves.

It is deliberately not a volume auto-applier. A posting doesn't get turned into a packet unless it's confirmed live in a real browser, confirmed recently posted, and its actual application questions have been captured off the live form — not guessed. If any of that fails, the posting is skipped and logged, not silently downgraded into a best-effort packet.