JobShield AI — Hybrid Fake Job Detection System

An AI-powered platform that detects fraudulent job postings using a hybrid approach combining rule-based detection, NLP feature extraction, structured metadata analysis, a logistic-regression-style ML scorer, and explainable AI.

Stack: React + Vite + TypeScript (frontend) · Django + Django REST Framework + PostgreSQL (backend) · JWT authentication.

The frontend calls the Django REST API for authentication, job analysis, prediction storage, and history/analytics — there is no third-party BaaS involved.

Project layout

job_AI-main/
├── src/              # React frontend
│   ├── lib/api.ts     # Fetch wrapper + JWT token handling
│   ├── lib/auth.tsx   # Auth context (register/login/logout/profile)
│   ├── lib/data.ts    # Calls to /analyze, /predictions, /history, /dashboard
│   └── ml/            # Client-side rule engine + scorer (instant UI feedback)
└── backend/          # Django REST API
    ├── api/           # Models, views, serializers, urls
    ├── ml/            # Server-side rule engine + scorer (authoritative, persisted)
    └── jobshield_backend/settings.py

Job analysis runs twice by design: instantly in the browser for responsive UI, and again on the Django server, whose result is authoritative and is what gets saved to PostgreSQL and shown in History/Dashboard/Analytics.

1. Backend setup

Install PostgreSQL and create the database

psql -U postgres

CREATE DATABASE jobshield;
\q

Set up the Python environment

cd backend
python -m venv venv
venv\Scripts\activate        # Windows
source venv/bin/activate     # Mac/Linux

pip install -r requirements.txt

Configure environment variables

copy .env.example .env       # Windows
cp .env.example .env         # Mac/Linux

Edit .env and set DB_PASSWORD to your local PostgreSQL password (adjust DB_USER/DB_NAME too if you changed them).

Run migrations and start the server

python manage.py migrate
python manage.py runserver

Backend is now live at http://localhost:8000. Verify:

• http://localhost:8000/admin/ — should show a login page
• http://localhost:8000/api/ — should show a 404 listing all valid routes (expected, there's no route at the bare path)

Optional — create a superuser for Django admin access:

python manage.py createsuperuser

2. Frontend setup

Open a new terminal (keep the backend running):

cd job_AI-main
copy .env.example .env       # Windows
cp .env.example .env         # Mac/Linux
npm install
npm run dev

Frontend is live at http://localhost:5173. The default .env already points VITE_API_URL at http://localhost:8000/api, matching the backend above.

3. Using the app

1. Go to http://localhost:5173, sign up with an email + password (this creates a Django User + Profile row).
2. Log in, then use Analyze to paste/enter a job posting — you'll get an instant risk score, explanation, and rule breakdown.
3. Check History, Dashboard, and Analytics — these read from PostgreSQL via the Django API, so anything you analyze will show up there.
4. Settings lets you update your profile and change your password.

Docker alternative (backend only)

If you have Docker Desktop installed:

cd backend
cp .env.example .env
docker-compose up --build

This starts PostgreSQL and Django together, running migrations automatically.

Notes

• The frontend also computes a full prediction client-side (src/ml/) so the UI feels instant; the backend's response overwrites the displayed scores once it returns, since that's the version actually stored in Postgres.
• Auth uses JWT (djangorestframework-simplejwt); tokens are stored in localStorage and refreshed automatically on expiry.
• CORS is pre-configured in backend/.env.example to allow localhost:5173.