AutoApply.AI

An AI agent that rewrites your resume, finds matching jobs, and applies for them autonomously.

---

What It Does

Feature	Description
Resume optimization	Upload a PDF resume + job description → get an ATS-friendly tailored resume (PDF) or improvement suggestions (JSON).
Document parsing	Upload resume, cover letter, and Q&A sheet (PDF/Word) → parsed into structured data and stored in the extension.
Personalized answers	Backend generates tailored answers to application questions using your resume and cover letter.
Job search	Search jobs by keywords via Playwright (e.g. LinkedIn); requires credentials and can hit timeouts.
Browser extension	Detects job application forms, maps fields, and autofills from your stored data with human-like typing.

---

Project Structure

AutoApply.AI/
├── app/                    # FastAPI backend
│   ├── api/
│   │   ├── resume.py       # /optimize-resume, /get-resume-suggestions
│   │   ├── user_data.py   # /parse-user-documents, /generate-personalized-answers
│   │   └── jobs.py        # /search-jobs
│   ├── main.py
│   └── schemas.py
├── extension/              # Chrome/Edge extension (Manifest v3)
│   ├── content.js         # Form detection, field mapping, autofill
│   ├── popup.js / popup.html / popup.css
│   ├── background.js
│   ├── storage.js         # Chrome storage for resume/cover/QA
│   ├── fieldMapping.js    # Field name → type mappings
│   └── manifest.json
├── modules/                # Core logic (interfaces + implementations)
│   ├── implementations/
│   │   ├── langchain_resume_optimizer.py
│   │   ├── pypdf_processor.py
│   │   ├── html_pdf_generator.py
│   │   └── playwright_job_scraper.py
│   └── interfaces/
├── tests/
│   ├── integration/       # test_resume_api.py
│   └── unit/              # test_pdf_processor, test_resume_optimizer
├── templates/             # Resume HTML template
├── static/
├── run_server.py          # Start backend (uvicorn)
└── pyproject.toml         # Poetry dependencies

---

Quick Start

Backend

1. Install dependencies (Poetry):

   poetry install

2. Environment (optional):

   • Create a .env with OPENAI_API_KEY for resume/LLM features.
   • For job search: LINKEDIN_EMAIL, LINKEDIN_PASSWORD.

3. Run the API:

   poetry run python run_server.py

   API: http://127.0.0.1:8000 — docs at http://127.0.0.1:8000/docs.

Extension

1. Open chrome://extensions/ (or edge://extensions/).
2. Enable Developer mode → Load unpacked → select the extension folder.
3. Pin the extension; ensure the backend is running at http://localhost:8000 for upload/parse and personalized answers.

Detailed setup and debugging: see extension/QUICK_START_GUIDE.md and extension/LOCAL_RUN_AND_DEBUG.md.

---

What’s Working

• Backend
  • FastAPI app with CORS for the extension and local dev.
  • Resume: POST /api/optimize-resume (PDF in → tailored PDF out), POST /api/get-resume-suggestions (PDF + job desc → JSON suggestions).
  • User data: POST /api/parse-user-documents (resume/cover/QA PDF or Word → structured JSON); POST /api/generate-personalized-answers (resume + cover + questions → tailored answers).
  • Jobs: POST /api/search-jobs (Playwright-based search; depends on env and target site).
• Extension
  • Popup: upload/download/update for resume, cover letter, and Q&A sheet; status and dates; demographics and work-auth preferences in storage.
  • Storage: storage.js with get/save/download for resume, cover letter, QA; demographics and work-auth.
  • Form detection: platform detection (Workday, Greenhouse, Lever, LinkedIn, Indeed, generic); field identification via labels, placeholders, proximity; visual indicator and field highlights.
  • Autofill: data mapping from resume/cover/QA, typing simulation, delays, event dispatch; text/select/radio/checkbox; date/phone formatting; preview and fill-status UI; undo; optional personalized answers via backend; budget protection and manual-upload fallback for file inputs.
• Core modules
  • PDF text extraction (PyPDF), resume optimization and personalized answers (LangChain/OpenAI), HTML→PDF resume generation, Playwright job scraper interface and implementation.
• Tests
  • Unit tests for PDF processor and resume optimizer; integration tests for resume API (see note below in “Needs work”).

---

What Needs Work

• Job search
  • Requires LinkedIn (or other) credentials and is sensitive to login/UI changes; timeouts and general errors can occur. Robustness and error handling need improvement.
• Extension / autofill
  • File uploads: Programmatic file input is restricted by the browser; extension uses fallbacks (e.g. manual upload notification + download link). File upload handling could be refined per platform.
  • Forms: Multi-step and conditional fields, rich text editors, and complex dropdowns may not be fully handled. More platforms and field mappings can be added.
  • Validation: Some sites enforce formats (e.g. phone (XXX) XXX-XXXX); format detection and retry logic could be improved.
• Tests
  • tests/integration/test_resume_api.py: test_optimize_resume_success expects a JSON response, but POST /api/optimize-resume returns a PDF. Tests should target the correct endpoint (e.g. /get-resume-suggestions for JSON) or assert on PDF response.
• Docs and ops
  • README and extension guides are in place; API docs are in OpenAPI at /docs. Need to add a short “Troubleshooting” section (e.g. CORS, extension host permissions, .env).

---

API Summary

Method	Endpoint	Purpose
POST	/api/optimize-resume	Resume PDF + job description → tailored resume PDF.
POST	/api/get-resume-suggestions	Resume PDF + job description → JSON suggestions.
POST	/api/parse-user-documents	Resume / cover letter / QA (PDF or Word) → structured JSON.
POST	/api/generate-personalized-answers	Resume + cover + questions (+ optional job context) → personalized answers JSON.
POST	/api/search-jobs	Job search criteria → list of job listings (e.g. via Playwright).

---

Extension architecture

The extension is a Chrome/Edge Manifest v3 add-on with three main parts. Popup (popup.html + popup.js) lets users upload resume, cover letter, and Q&A documents; it talks to the backend to parse files and stores the result in Chrome local storage via storage.js. Content script (content.js + fieldMapping.js) runs on every page: it detects job-application platforms (Workday, Greenhouse, Lever, LinkedIn, Indeed, or generic), finds form fields using labels and placeholders, and shows a floating “Fill form” control. When the user triggers autofill, the content script maps stored data to fields using fieldMapping.js, optionally calls the backend for personalized answers to open-ended questions, then fills inputs with human-like typing and events. Background (background.js) is a service worker that relays messages and logs form-detection events. The extension expects the backend at http://localhost:8000 for parsing and personalized-answer APIs.

---

License

MIT.