Overview

The system enables students and tutors to ask natural-language questions that are answered using textbook content through a Retrieval-Augmented Generation (RAG) approach.

Architecture summary

The system consists of core services (ML and storage layers) reused by two main flows:

• An ingestion flow for indexing textbook content
• An query flow for real-time question answering

List of components:

• Backend API
• Embedding model
• LLM
• Object storage
• Vector database
• PostgreSQL

Ingestion Flow

1. Textbooks are collected from trusted sources and processed by a crawler or loader
2. Content is chunked into semantically meaningful units (e.g., sections or pages)
3. Each chunk is embedded using an embedding model