AutoApply AI

Upload your resume once. Let AI fill every job application for you.

---

What It Does

AutoApply AI is a Chrome extension that reads your PDF resume, builds a structured profile from it using AI, and then automatically fills job application forms on any job portal — LinkedIn, Greenhouse, Lever, Workday, and more.

You upload your resume once. After that, every application form is filled with a single click.

---

How It Works

Step	What Happens
1. Upload Resume PDF	You upload your PDF resume inside the extension popup
2. AI Extracts Your Profile	PyMuPDF extracts text from the PDF; DeepSeek AI parses it into structured fields — name, email, phone, skills, experience, education, LinkedIn, summary
3. Navigate to a Job Application	Open any job posting and click Apply. The content script automatically scans all form fields on the page
4. AI Maps Fields to Your Profile	DeepSeek AI reads the form's structure and maps each detected field to the correct value from your profile
5. Review and Confirm	A review screen shows every mapped value. Edit anything before submitting
6. Form Auto-Filled	One click fills every field with native event dispatching, compatible with React, Vue, and Angular forms

---

Architecture

Layer	Technology	Role
Chrome Extension	React 18 + Vite + Tailwind CSS + Manifest V3	Popup UI, form scanning, field injection
Backend API	Python + FastAPI + Uvicorn	Auth, profile storage, AI orchestration
Resume Parser	PyMuPDF (fitz)	Extracts raw text from uploaded PDF
AI Engine	DeepSeek API (deepseek-chat)	Profile extraction from resume + form field mapping
Database	MongoDB Atlas via Motor	Stores user accounts and parsed profiles

---

Features

• Resume-to-Profile — Upload a PDF; AI extracts all relevant fields automatically. No manual data entry
• Intelligent Field Mapping — DeepSeek AI reads the visible form structure and maps fields correctly, even for custom or unusual field names
• Custom Memory — Fields you fill manually on forms are remembered and used in future applications
• Review Before Fill — Every mapped value is shown on a review screen before anything is written
• One-Click Fill — Injects values into inputs, selects, and textareas with native event dispatch (works with React, Vue, Angular)
• Portal Detection — Automatically detects the ATS platform (Greenhouse, Lever, Workday, LinkedIn, etc.) and adjusts mapping logic
• Profile Chat Editor — After setup, refine your profile via a chat interface ("add Python to my skills", "update my job title")
• Persistent Auth — JWT-based login with 365-day sessions stored in chrome.storage.local
• Dark Glass-Morphism UI — Polished dark popup with smooth animations

---

Supported Job Platforms

Platform	Detection
LinkedIn Easy Apply	Automatic
Greenhouse	Automatic
Lever	Automatic
Workday	Automatic
Indeed	Automatic
iCIMS	Automatic
Taleo	Automatic
SmartRecruiters	Automatic
Ashby	Automatic
BambooHR	Automatic
Any HTML form	Generic fallback

---

Project Structure

AutoApply/
├── backend/
│   ├── main.py              # All API routes: auth, profile, resume parsing, AI mapping
│   ├── models.py            # Pydantic models for all request/response schemas
│   ├── auth.py              # JWT signing, bcrypt hashing, token verification middleware
│   ├── database.py          # Async MongoDB connection using Motor
│   ├── requirements.txt     # Python dependencies
│   └── .env                 # Environment variables (not committed)
│
└── extension/
    ├── public/
    │   ├── manifest.json    # Chrome Manifest V3 config
    │   ├── background.js    # Service worker — tab event listener
    │   ├── content.js       # Form scanner: detects portals, extracts fields, fills inputs
    │   └── icons/           # Extension icons (16, 48, 128px)
    └── src/
        ├── components/
        │   ├── Auth/        # Login and Register screens
        │   ├── Onboarding/  # ResumeUpload — the primary onboarding flow
        │   ├── Chat/        # ProfileChatEditor — post-setup profile editing via chat
        │   ├── Review/      # ReviewScreen and FieldCard — confirm before fill
        │   └── UI/          # Button, Spinner, Toast
        ├── hooks/
        │   ├── useFormFill  # Orchestrates scan → map → fill pipeline
        │   └── useAuth      # Auth state and logout
        ├── services/
        │   ├── api.js       # HTTP client for all backend calls
        │   └── storage.js   # chrome.storage helpers (token, user, resume)
        └── pages/
            ├── PopupApp.jsx # Root component — routes between views
            └── Dashboard.jsx # Main dashboard — fill button, profile view, resume re-upload

---

Setup and Installation

Prerequisites

• Python 3.10 or higher
• Node.js 18 or higher
• A MongoDB Atlas cluster (free tier is sufficient)
• A DeepSeek API key

---

1. Backend

cd backend

# Create and activate a virtual environment
python -m venv venv
venv\Scripts\activate          # Windows
# source venv/bin/activate     # macOS / Linux

# Install dependencies
pip install -r requirements.txt

Create backend/.env:

MONGODB_URI=mongodb+srv://<username>:<password>@cluster.mongodb.net/autoapply
DEEPSEEK_API_KEY=your_deepseek_api_key_here
JWT_SECRET=a_long_random_secret_string
PORT=8000

Start the server:

uvicorn main:app --reload --host 0.0.0.0 --port 8000

API runs at http://localhost:8000. Interactive docs at http://localhost:8000/docs.

---

2. Extension Build

cd extension
npm install
npm run build

This outputs the packaged extension to extension/dist/.

---

3. Load in Chrome

1. Go to chrome://extensions
2. Enable Developer Mode (top-right toggle)
3. Click Load unpacked
4. Select the extension/dist/ folder
5. Click the AutoApply icon in your toolbar

---

API Reference

Method	Endpoint	Auth	Description
POST	/api/auth/register	No	Create a new account
POST	/api/auth/login	No	Login and receive a JWT
GET	/api/auth/me	Yes	Get current user info
GET	/api/profile	Yes	Fetch saved profile
PUT	/api/profile	Yes	Overwrite profile fields
POST	/api/profile/parse-resume	Yes	Upload PDF → AI extracts profile
POST	/api/profile/chat-update	Yes	Update profile via natural language
POST	/api/profile/memory	Yes	Save custom field answers for future use
POST	/api/ai/map-fields	Yes	Map detected form fields to profile data

Protected endpoints require Authorization: Bearer <token> header.

---

Environment Variables

Variable	Required	Description
MONGODB_URI	Yes	MongoDB Atlas connection string
DEEPSEEK_API_KEY	Yes	DeepSeek platform API key
JWT_SECRET	Yes	Secret used to sign and verify JWT tokens
PORT	No	Server port, defaults to 8000

---

Tech Stack

Component	Technology
Extension Frontend	React 18, Tailwind CSS v3, Vite 5
Chrome Extension API	Manifest V3, Content Scripts, Service Worker
Backend	FastAPI, Uvicorn (ASGI)
Database	MongoDB Atlas, Motor (async driver)
Authentication	PyJWT, bcrypt (passlib)
Resume Parsing	PyMuPDF (fitz) — PDF text extraction
AI Integration	DeepSeek API via OpenAI-compatible client
Form Compatibility	Native input/change event dispatch — React, Vue, Angular

---

Development Notes

• The backend must be running locally for the extension to work
• For production, update API_URL in extension/src/services/api.js to your hosted server URL
• Resume upload supports PDF only (max 5MB). The first 15,000 characters of extracted text are sent to the AI
• The content script builds a "form outline" — visible page text with embedded field markers — which gives the AI full context about what each field is asking
• Custom memory: when a user manually edits a mapped field and confirms, that label-value pair is saved and used automatically in future applications

---

Contributing

1. Fork this repository
2. Create a feature branch: git checkout -b feature/your-feature-name
3. Commit your changes with descriptive messages
4. Push to your fork and open a Pull Request

---

License

This project is licensed under the MIT License.

---

Built by Pritam Mundhe