Resumer — ATS Resume Builder
A web app that turns raw candidate inputs into a modern, recruiter-approved,
metric-driven resume optimized for Applicant Tracking Systems (ATS).
It works in two modes:
- AI-enhanced — when an
ANTHROPIC_API_KEY is configured, a Vercel serverless
function calls the Claude API to rewrite bullets into natural XYZ-standard
achievements and craft the professional summary. The key stays server-side.
- Rule-based — with no key configured, a deterministic in-browser engine handles
scoring, keyword matching, and heuristic bullet rewriting. The app falls back to this
automatically, so it always works, even opened as a local file.
The active mode is shown as a badge on the diagnostic panel.
What it does
Given a candidate's target role, raw work history, and skills, Resumer produces the three
output modules of the resume-building engine:
-
ATS & Match Score Diagnostic
- Estimated ATS compatibility score (0–100) from weighted signals.
- Real-time keyword matching against a pasted job description (matched vs. missing).
- Readability & impact grade.
-
Complete Modern Resume (copy-paste ready)
- Clean, single-column, parser-safe markdown.
- Standardized headers: Professional Summary, Core Competencies, Professional Experience,
Key Projects, Education & Certifications.
- No tables, text boxes, icons, or characters that break ATS parsers.
-
Resume Optimization Audit
- "Before & After" bullet transformations using the Google XYZ standard
— "Accomplished [X], as measured by [Y], by doing [Z]".
- Recommendations for high-value additions (certifications, portfolio links, quantification).
Engine logic
- ATS scoring weights keyword coverage (30), quantified achievements (25),
action-verb openers (20), core-field completeness (15), and parser-safe structure (10).
- Keyword matching extracts frequency-ranked words and phrases from the job description
and checks them against the candidate corpus.
- Bullet analysis grades each line for metrics + action verbs, then rewrites weak,
responsibility-driven statements into quantified achievements.
- Skill structuring separates technical skills from strategic/functional capabilities.
See js/engine.js — the engine is pure and framework-free.
Run it locally
Rule-based only (no key), zero setup:
python3 -m http.server 8000 # then open http://localhost:8000
With the AI backend (runs the serverless function locally):
npm i -g vercel
cp .env.example .env.local # add your ANTHROPIC_API_KEY
vercel dev # serves the site + /api/enhance
Click Load sample to see a full end-to-end example.
Deploy to Vercel
- Push this repo to GitHub (already done if you're reading this on
main).
- In Vercel: Add New → Project → Import this repo. It's detected as a static
site with serverless functions in
api/ — no build command needed.
- Add an environment variable
ANTHROPIC_API_KEY (Project → Settings →
Environment Variables) to enable AI mode. Without it, the app deploys fine and
runs in rule-based mode.
- Deploy. Or from the CLI:
vercel (preview) / vercel --prod.
Test the engine
npm test # node test/engine.test.js
Project structure
index.html # UI
css/styles.css # styles (light/dark aware, print-friendly)
js/engine.js # pure ATS engine (scoring, keywords, generation, audit)
js/app.js # UI controller (form ↔ engine + AI backend, copy/download)
api/enhance.js # Vercel serverless function — calls the Claude API
vercel.json # Vercel config
test/engine.test.js # lightweight assertions for the engine
Export
- Copy markdown — paste anywhere.
- Download .md — the raw markdown file.
- Download PDF — opens a print-optimized, single-column view and triggers the
browser's print-to-PDF. Text stays selectable (ATS-parseable), not rasterized.
Allow pop-ups for this button.
Roadmap
- DOCX export.
- Multiple experience entries and templates.
- Semantic (embedding-based) keyword matching.
