Exam Review Bot
Published:
Overview
Exam Review Bot (ERB) is a custom intelligent retrieval and hybrid RAG system for final exam preparation. The system combines document processing, vector search, and LLM-powered responses to help students efficiently review course materials.
Key Features
- Document Processing: Extracts specific lecture PDFs from your course schedule
- Intelligent Matching: Matches documents to user questions using semantic search
- Knowledge Augmentation: Supplements with internet knowledge through external search
- Smart Summarization: Generates intelligent summaries and explanations
Technical Stack
- Backend: FastAPI, Python
- Frontend: React, TypeScript, Vite
- AI/ML: OpenAI API, Langchain, ChromaDB
- Vector Store: FAISS/Chroma for efficient similarity search
Architecture
The system follows a modular RAG pipeline:
- Document Ingestion: PDF loader and text chunker process course materials
- Embedding Layer: Creates vector embeddings for semantic search
- Vector Store: Stores embeddings in ChromaDB for fast retrieval
- Retriever: Searches the database for relevant context
- LLM Integration: OpenAI/Claude models generate contextual answers
Features in Development
- Custom indexing and vectorstore database
- File upload interface enhancements
- Progress tracking
- User authentication
- Interactive demo integration
Links
- GitHub Repository
- Topics: JavaScript, CSS, Python, TypeScript, Hybrid RAG