Skip to main content

About the project

Zimbabwe Nursing Tutor is an AI study tutor for Zimbabwean nursing students. Instead of answering from a chatbot’s general knowledge, it answers only from approved Zimbabwean guidelines (the Essential Medicines List, national HIV and postnatal care guidelines) and shows the exact passage, page and edition behind every point.

Each question goes through a safety check (patient identifiers, emergencies, patient-specific dosing), a search of the guideline text using Gemini embeddings, and an evidence gate. If the sources don’t cover the question, Gemini is never called and the app says so. Otherwise Gemini writes a structured, cited answer, and our code then verifies every quote and number against the source text before showing it. Answers that fail the check are withheld.

Students can learn in 11 modes (explain, step-by-step, memory aid, quiz, exam style, flashcards and more), tick off each point on a visual learning map, and listen to a friendly read-aloud voice. It is educational support only, and it shows each source’s edition and currency status.

Built with

Python, Streamlit, Google Gemini API (google-genai SDK): Gemini embeddings for semantic search, Gemini structured JSON output for cited answers, Gemini text-to-speech for the read-aloud voice; PyMuPDF (PDF processing), NumPy, BM25 keyword search, python-dotenv, Git and GitHub.

Prize Categories Entered

  • Google

    Best Use of the Google Gemini API

    Submitted