Selected work.

A tool-augmented LLM system for iterative student feedback in structured writing tasks. Built to explore practical AI applications in an educational context — specifically, how AI can be scaffolded to give substantive, personalised feedback without replacing the judgement of a human teacher.

The system chains multiple prompts with retrieval from a rubric document, generates critique at multiple levels (argument structure, evidence use, expression), and iterates based on student responses. Python backend, FastAPI endpoints, minimal front-end.