Building Intelligent Educational Agents: A Scalable LLM-Based Framework for Assessment Generation

Abstract

With scalable and adaptive learning solutions being more important than ever before, this paper introduces an AI-powered framework, built on the ClassQuiz platform, that revolutionizes educational assessment. It leverages Large Language Models (LLMs), prompt engineering, and retrieval-augmented generation (RAG) to automatically create high-quality, contextually relevant learning materials. The solution focuses on developing LLM capabilities for content generation, creating LLM-driven educational agents, and providing a stable, production-ready system with distributed processing and API integration. This framework aims to deliver personalized, scalable for large-scale use through the implementation of parallelized data processing, and inclusive learning experiences by offering confgurable assessment templates, semantic content interpretation, and adaptive learning paths, ultimately enriching both content creation for educators and the learning journey for students. It is adaptable to various teaching techniques, as it leverages context and instructions provided by teachers and students.