A resume as code
My resume hadn’t been touched in years, and when I finally sat down to fix it I made the same mistake most engineers make: I opened a word processor. Twenty minutes in, I realized I was hand-formatting the same work history four times — once per version I wanted — and that this is exactly the problem we solve at work every day. Repetition with small variations isn’t an editing problem. It’s a build problem.
So the resume became a repo.
The shape
Four Markdown files are the source of truth, one per audience: a core DevOps/platform version, an AI-infrastructure version, an SRE version, and a neutral chronological one. They share the same history — the bullets and emphasis shift per audience. Everything else is generated:
- HTML with a small print stylesheet — open in a browser, print to PDF.
- Word for the recruiters and applicant tracking systems that want .docx.
- LaTeX for when typesetting matters, built on Overleaf.
One build.sh and pandoc do all of it. Edit a bullet in one Markdown file, run the
script, and every format is current. The repo is private; the outputs are committed so
each version is one click away.
What it wasn’t
This wasn’t about the tooling being clever — pandoc has done Markdown-to-anything for
years. It was about applying the discipline we already trust: a single source of truth,
diffs instead of guessing what changed, and review before anything ships. When a recruiter
asks “what changed since the version you sent in March,” git log answers.
Two deliberate choices worth stealing:
Placeholders for contact info. The committed sources say [YOUR-EMAIL] and
[YOUR-PHONE]. Personal contact details don’t belong in git history, even private git
history — repos get transferred, cloned, and shared more casually than documents do. Fill
them in at send time.
Comments as TODO markers. The eras of my career I couldn’t document well yet carry an HTML comment in the source — invisible in every build, loud in every edit. The resume ships while the backlog stays honest.
The payoff
Four tailored resumes that stay in sync cost me one afternoon to set up and now cost nothing to maintain. The next update is a pull request against my own history — reviewed, built, and versioned like everything else I ship.