Qesem — AI Study Assistant

Table of Contents

• Overview
• Features
• Tech Stack
• Architecture
• Getting Started
• Configuration
• Running the Apps
• API Endpoints
• Usage Examples
• Folder Structure
• Contribution
• License

Overview

Qesem is an AI-powered study assistant. Students can upload PDFs or text notes, chat with an agentic assistant, generate quizzes automatically, and receive graded results. Responses stream in real time via Server-Sent Events (SSE), and answers are grounded with Retrieval-Augmented Generation (RAG) over your uploaded materials.

Features

• RAG over uploads: Chunking and vector embeddings (Voyage AI) stored in MongoDB for grounded responses.
• Agentic flows: LangChain + LangGraph for quiz generation, grading, and contextual reasoning.
• Realtime SSE: Streaming assistant responses for smooth UX.
• Quiz automation: Auto-generate quizzes and grade user-submitted answers.
• File uploads: PDF/TXT ingestion with size limits and MIME validation.
• Session history: Persisted conversations with resume capability.
• Configurable LLMs: Gemini model selection via env; Voyage embed model override.
• Dark UI: React + Vite + Tailwind CSS 4 with responsive layout.

Tech Stack

• Frontend: React 19, TypeScript, Vite, Tailwind CSS 4, Zustand.
• Backend: Node.js, Express 5, TypeScript, MongoDB (Mongoose), SSE.
• AI/ML: LangChain, LangGraph, Gemini (Google Generative Language API), Voyage embeddings.
• Build/Tooling: Vite, ESLint, ts-node, TypeScript.

Architecture

• client/: Vite React SPA; chat UI, quiz UI, SSE client, upload trigger.
• server/: Express API; routes for chat, quiz, upload, sessions, SSE; LLM + embedding services; MongoDB persistence.

Getting Started

Prerequisites

• Node.js 18+ (recommended)
• npm 9+
• MongoDB Atlas or local MongoDB
• API keys: Google Generative Language (Gemini) and Voyage AI

Clone & Install

git clone https://github.com/CSEC-ASTU/Qesem.git
cd Qesem
npm install --prefix server
npm install --prefix client