Skip to content
ATS Screener

How It Works

ATS Screener combines client-side document parsing with AI-powered analysis to simulate how each platform would evaluate your resume.

Architecture

Section titled “Architecture”
Browser                          Server (Edge Functions)
┌─────────────────────┐          ┌──────────────────────┐
│ 1. Upload PDF/DOCX  │          │                      │
│ 2. Parse in Worker   │──text──▶│ 3. Gemini AI scores  │
│ 4. Display results  │◀─JSON───│    from 6 platforms   │
└─────────────────────┘          └──────────────────────┘

Step 1: Client-Side Parsing

Section titled “Step 1: Client-Side Parsing”

When you upload a resume:

  • PDF files are parsed using Mozilla’s pdfjs-dist running in a Web Worker. This extracts all text content, preserving structure.
  • DOCX files are parsed using mammoth, which converts Word documents to clean text.

All parsing happens in your browser. The file itself is never uploaded anywhere.

Step 2: Text Extraction

Section titled “Step 2: Text Extraction”

The parser extracts structured data:

  • Contact info (name, email, phone, LinkedIn, GitHub)
  • Sections (experience, education, skills, projects, certifications, summary)
  • Dates and timeline information
  • Raw text for keyword analysis

Step 3: AI Analysis

Section titled “Step 3: AI Analysis”

The extracted text (not the file) is sent to a SvelteKit server endpoint, which proxies it to Gemini AI. The AI evaluates the resume from 6 different platform perspectives, each with unique:

  • Parsing strictness levels
  • Keyword matching strategies (exact, fuzzy, semantic)
  • Section weight distributions
  • Scoring quirks and edge cases

Step 4: Results

Section titled “Step 4: Results”

The AI returns structured JSON with per-platform scores, breakdowns, keyword analysis, and suggestions. The frontend renders this as an interactive dashboard.

Privacy

Section titled “Privacy”

Your privacy is protected at every step:

  1. Files stay local. PDF/DOCX parsing happens entirely in your browser’s Web Worker. The original file is never uploaded.
  2. Only text is transmitted. The extracted text (not the original file) is sent to Google Gemini for AI analysis.
  3. Scan history is stored in Firebase. Your scores and scan metadata are saved to your account in Firestore so you can track progress over time. Only you can access your own data.
  4. Free account required. Sign in with Google or email/password to use the scanner. This enables scan history and prevents abuse.
  5. Open source. You can verify all of the above by reading the source code.