Building an AI Tutor that Learns with Students: A Voice-Driven, Document-Aware. Step-by-Step Education Assistant

Abstract

This thesis presents the design and development of an AI-powered mobile educational assistant aimed at improving learning accessibility for students in African communities. Built using Flutter, the app integrates advanced AI technologies such as large language models, speech recognition, text-to-speech, and document analysis into a unified platform. It supports over 9 languages, including several African languages, enabling students to learn in their native dialects. The system uses a FastAPI backend with Supabase for data management and includes AI-driven quizzes to assess and track student progress. Overall, the project demonstrates that modern AI APIs and free cloud infrastructure can effectively deliver a scalable and accessible intelligent tutoring system.