TeachPro: A Unified Desktop Application for Intelligent Classroom Management and LLM-Powered Question Generation

Teachers in contemporary institutions handle attendance, assessment, question preparation, and student data through disconnected tools, which increases administrative workload and reduces instructional focus. In this work, we developed TeachPro, a cross-platform desktop application built with MongoDB, Express.js, React.js, Node.js, and Electron.js that unifies eight teaching workflows and integrates a large language model pipeline (Google Gemini 1.5 Pro) for question and quiz generation. The system includes student management, notes, question bank, rich-text exam paper authoring with AI insertion, shareable AI quiz delivery, attendance analytics, timetable management, and secure JWT-based access. We present the architecture, implementation details, prompt-engineering strategy, and an evaluation protocol combining performance profiling and teacher pilot feedback. Across 100 generated questions, factual accuracy reached 84–96% across difficulty levels, Bloom’s taxonomy alignment reached 81–89%, and language clarity remained above 95%. The platform was built and tested in a real academic environment with faculty users during routine assessment preparation. Pilot use indicated 60– 70% reduction in repetitive administrative effort and faster end-to-end quiz creation. This manuscript provides a practical, reproducible blueprint for applying generative AI in everyday classroom operations with teacher-in-the-loop control.