MzeeChakula — Model-Testing

This repository contains the frontend and backend for the MzeeChakula nutrition recommendation and meal-planning prototype. The project provides:

• An interactive, voice-enabled chat assistant for collecting user demographics and dietary preferences.
• A prediction API that estimates daily caloric needs and recommends local foods based on nutritional similarity and learned models.
• A small frontend app built with Vue + Vite + Pinia that consumes the backend API.

Live demos

• Frontend (static UI): https://mzeechakula.vercel.app/
• Backend API: https://mzeechakula-backend.onrender.com/
• Project Documentation: Documentation/README.md (Run locally for full site)

Quick links

• Frontend folder: frontend/
• Backend folder: backend/

Getting started (developer)

1. Clone the repository:

git clone https://github.com/MzeeChakula/Model-Testing.git
cd Model-Testing

2.Backend (Python / FastAPI)

• Requirements: Python 3.12+ and a virtual environment.

# from repo root
cd backend
python3 -m venv ../.venv
source ../.venv/bin/activate
pip install -r requirements.txt
# Run the API server (development)
uvicorn api.main:app --reload --host 127.0.0.1 --port 8000

Notes:

• The backend serves endpoints under /predict, /foods, and /health.
• Model files and optional embeddings are expected under backend/models/. If no trained models are present the backend falls back to a heuristic predictor so the API remains usable.

3.Frontend (Vue / Vite)

• Requirements: Node.js + npm (or yarn).

# from repo root
cd frontend
npm install
# Run dev server
npm run dev
# Build for production
npm run build

• The frontend expects an API base URL in VITE_API_URL (defaults to http://localhost:8000). When running locally, start the backend first.

Recommendations & troubleshooting

• If the recommender system (embedding-based) is not available, the frontend falls back to scoring local foods by basic nutritional similarity; ensure frontend/src/store/predictionStore.js has been loaded and that getLocalFoods works (endpoint: /foods/local).
• To include the built frontend in this repo for hosts like Vercel that can serve static output, the dist/ folder can be committed — the project .gitignore has been updated to allow committing dist.

Deploying

• Frontend: Vercel (recommended). Point the Vercel project to the frontend directory or push the built dist folder to the repo and configure Vercel to serve it.
• Backend: Render / Heroku / any container or VM. Ensure environment variables and model files are available on the host.

Useful git commands (commit dist to repo)

# Force add a previously ignored folder (only if needed)
git add -f frontend/dist
git commit -m "Include built frontend dist for deployment"
git push origin main

Contributing

• Open issues and PRs are welcome. Keep changes focused and add tests where appropriate.

License

• This repository is released under the terms in the LICENSE file.

Contact

• For questions, open an issue or contact the repository owner.