AI-Assisted Student Testing Platform

Situation

K-12 student assessments often focus only on final scores, providing limited insight into how students approach questions or where they struggle during a test. Traditional testing platforms rarely capture granular interaction data that could be used to evaluate student learning patterns, question quality, or overall test fairness.

Task

My responsibility was to design and build a web based testing platform for K-12 students and then recreate the same application using ChatGPT to evaluate the capabilities and limitations of large language models in building functional, data driven web applications for educational use cases.

Action

• Designed and developed a web based testing website that allowed students to take assessments in a structured and user friendly interface
• Implemented logging of detailed testing metadata, including test date and time, total time taken, questions attempted, and correct versus incorrect responses
• Captured per question response time to analyze student behavior, pacing, and difficulty levels at a granular level
• Structured data collection to support evaluation of student performance, question quality, and test fairness
• Rebuilt the same application using ChatGPT to assess how effectively an LLM could replicate functional requirements, data logging logic, and basic application architecture

Result

• Produced a functional assessment platform capable of generating actionable learning and performance metrics beyond final test scores
• Enabled deeper analysis of student behavior, including pacing, difficulty bottlenecks, and learning gaps
• Provided data-driven insights into test quality and fairness by analyzing question level response patterns
• Demonstrated practical strengths and limitations of LLMs in building basic, real world web applications for education