As team lead and sole full-stack developer, I led a group of seniors and graduate students in an upper-division AI/ML course at Mines in building the essay assistant shown on the top left. It combines prompt engineering, multi-threaded Python, GPT-4o Mini, regex, text search, response normalization, and more to generate targeted, rubric-based feedback. Our goal was to match the guidance level of a typical tutor, not do the work for the student, unlike most generative AI tools. While we were limited by time constraints, the AGILE development framework I had us working under allowed us to accomplish more than we originally anticipated. The upper-right panel next to the app is an addition I made for this website: a filtered real-time progress feed that visualizes each processing stage while preserving app security.

To the right is a flow-chart describing the overall system architecture in detail: