PathMind - AI College Mentor 🎓

PathMind is an AI-powered application designed to help first-generation college students and aspirants explore colleges, understand admission processes, and get instant answers to their questions using Google's Gemini AI.

The application is built with a focus on simplicity, responsiveness, and accessibility in multiple languages.

🚀 Tech Stack

• Frontend Framework: React 18
• Build Tool: Vite v8
• Styling: Tailwind CSS (via inline utility classes) & Vanilla CSS
• Icons: Lucide React
• AI Integration: Google Gemini API (gemini-2.5-flash)

📂 Key Files & Project Structure

To upload this project to GitHub, you only need the following important files and folders. Do NOT upload the node_modules folder or .env file containing your API key.

PathMind/
├── src/
│   ├── App.jsx        # 🌟 Main application file (Contains all UI, State & Logic)
│   ├── index.css      # Base Tailwind imports
│   └── main.jsx       # React DOM rendering entry point
├── public/            # Static assets (favicons, etc.)
├── index.html         # Main HTML template
├── package.json       # Project dependencies and scripts
├── vite.config.js     # Vite configuration
└── README.md          # This documentation file

🧠 Core Logic & Features

The entire application logic is consolidated into a single monolithic file (src/App.jsx) for rapid prototyping. Here is how the core systems work:

1. Two-Screen Architecture

The app switches between two main components based on the college state variable:

• HomeScreen: The landing page where users can search for a college or select a quick chip (e.g., Stanford, MIT).
• ResultScreen: The detailed view for a specific college that shows data snapshots, eligibility criteria, and a chat interface.

2. Multi-Language Support

The app supports 14 languages (English, Hindi, Spanish, French, Portuguese, Chinese, Arabic, German, Japanese, Russian, Korean, Italian, Tamil, Telugu, Marathi).

• A LANGS array holds language definitions and specific system instructions for the AI.
• A massive UI object holds translations for every static text element, button, and prompt on the screen.
• State (lang) is passed down to dynamically update the UI in real-time.

3. AI Integration (Gemini)

The app uses the fetch API to talk directly to generativelanguage.googleapis.com.

• Chat Interface: Users can ask free-form questions. The app injects a system prompt instructing the AI to act as a helpful college mentor, responding in the user's selected language.
• JSON Data Extraction: To populate the "College Snapshot" (World Rank, Estd, Rating, Accreditations) and Admission logic, the app prompts Gemini to return strict JSON.
• Robust Parsing: Since LLMs sometimes wrap JSON in markdown (json ... ), the askGemini helper uses a smart extraction method (indexOf('{') and lastIndexOf('}')) to guarantee the app doesn't crash even if the AI includes conversational text.

4. Safety Interceptor

The application includes a client-side keyword scanner. If a user inputs crisis-related keywords (e.g., self-harm, depression), the app intercepts the request before it hits the API and immediately renders a supportive message with emergency helpline information.

5. Future Self Feature

A unique UI element where users can "Talk to their future self". The AI is prompted to adopt the persona of a 28-year-old successful professional who graduated from the selected college, offering empathetic and encouraging advice.

🛠️ How to Run Locally

1. Install Dependencies:

   npm install

2. Setup API Key: Create a .env file in the root directory (do not upload this file to GitHub) and add your Gemini API key:

   VITE_GEMINI_API_KEY=your_api_key_here

3. Start Development Server:

   npm run dev

4. Build for Production:

   npm run build

🎨 Design Philosophy

• Mobile-First: The app uses a single-column layout centered on the screen, optimizing the experience for mobile devices where most first-gen students access information.
• Accessible UI: Heavy use of clear typography, subtle gradients (Mint Green), and iconography (Lucide) to reduce cognitive load.