Akeredolu-Samuel/oluko

Language-first, 100% offline educational assistant bridging Nigeria’s learning poverty and teacher capacity gap. # Oluko Oluko is an offline-friendly AI classroom companion for Nigerian students and teachers. It uses Ollama with a local Gemma model to power tutoring, lesson-plan generation, text simplification, and quiz creation through a simple FastAPI backend. ## What it does - Student tutoring in English and major Nigerian languages - Lesson plan generation for teachers - Text simplification for different class levels - Quiz generation in a WAEC-style format ## Project structure - main.py: FastAPI application exposing the tutoring endpoints - Modelfile: Ollama model definition for the local assistant - model/: local model assets - oluko_app/: Flutter mobile app frontend ## Offline setup 1. Install Ollama: ollama.com 2. Start the Ollama service locally. 3. Place your local model file in the project root as gemma-4-q4.gguf. 4. Build the model locally: ```bash ollama create oluko -f Modelfile ``` 5. Run the FastAPI server: ```bash python main.py ``` If you prefer to use uvicorn directly: ```bash uvicorn main:app --reload ``` ## Notes - Large model files such as gemma-4-q4.gguf are ignored by Git to keep the repository lightweight. - The app expects Ollama to be running locally at localhost.