WiBD GenAI Hackathon 2026

Team Name

Team Umang

---

Problem Statement

• Problem Statement Number: 1, 3, 4
• Problem Statement Title: Research Paper Summarizer, College Info Chatbot, Smart Study Companion

---

Project Overview

COLLEGE MATE is an AI-powered learning platform that combines three intelligent tools to enhance the student experience:

• What problem we are solving: Students struggle with research paper comprehension, finding college information, and creating effective study materials. Our platform addresses all three challenges in one unified solution.
• Who it is for: College students, researchers, and anyone seeking to streamline their academic journey.
• Why it is useful: It saves time by summarizing complex research papers, provides instant college-related answers, and generates personalized study materials including flashcards and quizzes.

---

Tech Stack

• Programming Language(s): TypeScript, Python
• Frameworks / Libraries: Next.js 14 (React), FastAPI, Framer Motion, Lucide React
• LLMs / APIs: Google Gemini API, ArXiv API
• Database / Vector Store: MongoDB
• Authentication: JWT, Google OAuth, GitHub OAuth
• Deployment: (To be added)

---

Architecture / Approach

Our solution follows a modern full-stack architecture:

1. Frontend: Next.js App Router with server actions for secure authentication
2. Backend: FastAPI microservices handling AI processing and external API integrations
3. AI Integration:
   • Research Summarizer uses ArXiv API for paper search and Gemini for summarization
   • College Chatbot uses RAG with document embeddings for context-aware responses
   • Study Companion generates explanations, flashcards, and quizzes using Gemini

---

Setup Instructions

1. Clone the repository

   git clone https://github.com/WIBD-Vadodara/Umang.git
   cd Umang

2. Frontend Setup

   cd fe
   npm install

3. Backend Setup

   cd be
   pip install -r requirements.txt

4. Environment Variables Create .env.local in /fe:

   MONGODB_URI=your_mongodb_uri
   JWT_SECRET=your_jwt_secret
   GOOGLE_CLIENT_ID=your_google_client_id
   GOOGLE_CLIENT_SECRET=your_google_client_secret
   GITHUB_CLIENT_ID=your_github_client_id
   GITHUB_CLIENT_SECRET=your_github_client_secret
   

5. Run the Application

   # Terminal 1 - Frontend
   cd fe && npm run dev
   
   # Terminal 2 - Backend
   cd be && uvicorn main:app --reload

---

Repository Structure

• /fe → Frontend (Next.js)
• /be → Backend (FastAPI)
• /docs → PPT, reports, architecture details
• /assets → Images, screenshots, diagrams

---

Team Members

• Yash Bharvada – Yash-Bharvada
• Kushal Desai - KushalvDesai
• Krish Devani - KrishDevani30
• Pankti Akbari - pankti0409
• Kathan Modh - KathanModh259

---

Notes / Assumptions

• Requires active internet connection for LLM API calls
• MongoDB Atlas recommended for database
• ArXiv API has rate limits for paper searches
• Future improvements: offline mode, more LLM options, mobile app

---

Submission Declaration

This project was developed as part of WiBD GenAI Hackathon 2026 and all code was written during the hackathon period.