AI / ML · Full-Stack Developer
QuestionBot
An end-to-end Python/Streamlit tool that converts veterinary (or any) lecture PDFs into structured, exam-ready question banks using the OpenAI API, with a full browser-based UI for generation, quizzing, and multi-format export.
Highlights
- 01
Automated generation of 25+ exam questions per topic from raw lecture PDFs by building a two-stage OpenAI pipeline (topic discovery → structured question generation) with strict JSON Schema validation, eliminating hours of manual question-writing per exam cycle.
- 02
Drove redundant API spend to near-zero by implementing MD5-based content-aware caching keyed on file hash + model + question count, skipping regeneration unless the source PDF or settings actually changed.
- 03
Shipped a 4-tab Streamlit UI (drag-and-drop PDF upload, per-topic browser, randomized interactive quiz with instant scoring, one-click export) supporting 4 export formats (JSON, Word .docx, Anki deck, CSV) across 3 question types (T/F, MCQ, multi-select).