Auto-Assess: Automated Evaluation Platform for Enhanced SQL learning

This work describes the designing and development of an AI-driven SQL learning and teaching platform that could be used for automating SQL query generation, evaluation, and feedback. Based on NLP and AI methodologies, the proposed platform makes use of Ollama’s Llama 3 for schema extraction, question generation, and grading. For answer generation, a fine-tuned Llama 3 model on a variation of the spider dataset is used. MongoDB has been utilized for scalable data storage and React for an intuitive front-end interface. The platform enables an instructor to upload datasets and automatically generate SQL-based tests for students. Students are also given immediate feedback regarding their SQL queries, which otherwise is a labor-intensive task. The platform can also overcome challenges that traditional SQL education face, such as inconsistency in grading and delay in giving feedback, thereby offering personalized feedback, comprehensive performance reports, and engaging learners. Our solution would lighten the burden of teaching SQL by ensuring less effort from instructors while teaching their students dynamically, through the assessment of queries in a real-time setting. By automation in conjunction with the feature of real-time interactivity, this solution has been steered toward an effective and scalable learning experience for SQL by both educators and learners.