ATS Resume Parser & Diagnostic Tool

A frontend-focused tool that shows what an Applicant Tracking System (ATS) actually sees when it parses your resume — and highlights issues that commonly cause data loss, keyword misses, or broken links.

This project focuses on diagnosis, transparency, and correctness, not resume rewriting or AI-driven auto-fixing.

---

Why This Project Exists

Many job portals (Workday, Oracle Cloud, Greenhouse, etc.) silently mis-parse resumes due to:

• Broken Unicode characters (e.g. Ɵ, Ʃ, ligatures)
• Token splits caused by PDF fonts (ex p anding, Tik T ok)
• Malformed URLs with invisible spacing
• Non-ASCII glyphs that look correct to humans but fail machine parsing

Most resume tools hide these failures.

This app makes them visible and explainable.

---

What This Tool Does

1. ATS-Style Text Extraction

• Parses PDF and DOCX resumes similarly to real ATS systems
• Displays the raw extracted text exactly as machines see it

2. ATS Diagnostics

Detects and highlights:

• Unicode ligatures and non-ASCII glyphs
• ATS-breaking characters inside keywords and URLs
• Token integrity issues (split words, broken hyphenation)
• URL formatting issues that break ATS link detection

Each issue is classified by severity.

---

3. ATS Compatibility Score (Approximate)

• Computes a heuristic ATS compatibility score
• Based on parsing confidence, not hiring likelihood
• Mirrors how ATS systems degrade confidence internally

Score bands:

• 90–100 → ATS-safe
• 75–89 → Minor ATS issues
• 60–74 → Moderate ATS risk
• < 60 → High ATS failure risk

---

4. ATS-Safe Suggestions (Preview Only)

• Shows before → after suggestions for detected issues
• No automatic mutation
• No rewriting
• User remains fully in control

---

5. Readable ATS Output

• Toggle between:
  • Raw ATS Text (exact machine output)
  • Formatted ATS View (presentation-only)
• Formatting improves readability without changing content
• Helps users visually audit their resume structure

---

6. Accessible & Mobile-Friendly UI

• Keyboard navigable
• Screen-reader friendly (ARIA roles, live regions)
• Mobile-optimized without affecting desktop UX
• Collapsible sections with animated accordions to reduce overload

---

Tech Stack

Frontend

• React
• TypeScript
• Tailwind CSS (Vite setup)

Backend

• Node.js
• Express

Parsing

• pdfjs-dist for PDF parsing
• mammoth for DOCX parsing

Deployment

• Frontend: Vercel (React + Vite)
• Backend: Render (Node.js + Express)

The frontend is deployed as a static Vite build on Vercel.
The backend runs on a full Node.js runtime on Render to support reliable PDF and DOCX parsing using pdfjs-dist and mammoth.

Notes

• The backend runs on Render’s free tier and may have a cold start delay (up to ~50s) after inactivity.
• This split deployment was chosen to ensure correct ATS-style parsing, which is not well supported in edge or serverless runtimes.

Environment Variables

The frontend connects to the backend via:

VITE_API_URL=<backend-url>

---

Design Principles

• Truth over polish: show what ATS actually sees
• No silent fixes: nothing is changed without user awareness
• Explainability first: every score and warning is traceable
• Frontend-first: UI clarity and usability matter as much as logic
• Accessibility is not optional

---

What This Tool Is NOT

• ❌ A resume writer
• ❌ A hiring probability predictor
• ❌ An AI keyword stuffer
• ❌ An auto-submit optimizer

It is a diagnostic and educational tool.

---

Running Locally

Frontend

cd frontend
npm install
npm run dev

Backend

cd backend
npm install
npm run dev

---

Future Improvements (Optional)

• Export ATS-safe text
• Section-wise ATS confidence
• Per-issue apply/reset controls
• Dark mode (UI only)

---

Motivation

This project was built after repeatedly encountering resumes being mis-parsed by job portals despite looking correct visually.

The goal is simple:

If an ATS breaks your resume, you should be able to see why.