ChefAI

ChefAI is an iOS app that analyzes fridge contents from a photo and recommends recipes. Take a photo of your fridge, and the app identifies food items using GPT-4o's vision API, then suggests 3-5 recipes with difficulty ratings, calorie counts, ingredient lists, and step-by-step cooking instructions.

Tech Stack

• Frontend: React Native + Expo, TypeScript
• Navigation: Expo Router (file-based routing)
• Styling: NativeWind (Tailwind CSS for React Native)
• Async State: TanStack Query
• Validation: Zod
• Backend: Supabase (auth, storage, database)
• AI: OpenAI GPT-4o (vision + text)
• Camera: Expo ImagePicker

Prerequisites

• Node.js (v18 or later)
• Expo CLI (npm install -g expo-cli)
• Expo Go app on your iOS device (for testing on a physical device)
• An OpenAI API key
• A Supabase project (for storage)

Setup

1. Clone the repository:

   git clone https://github.com/GodSpeed0118/ChefAI.git
   cd ChefAI

2. Install dependencies:

   npm install

3. Configure environment variables:

   Create a .env file in the project root:

   EXPO_PUBLIC_SUPABASE_URL=your-supabase-url
   EXPO_PUBLIC_SUPABASE_ANON_KEY=your-supabase-anon-key
   EXPO_PUBLIC_OPENAI_API_KEY=your-openai-api-key
   

Running the App

# Start the Expo dev server
npx expo start

# Start with iOS simulator
npx expo start --ios

# Start in tunnel mode (for testing on a physical device over the network)
npx expo start --tunnel

To run on a physical device, scan the QR code shown in the terminal with the Expo Go app.

Project Structure

app/                  # Expo Router pages (file-based routing)
  _layout.tsx         # Root layout with providers
  index.tsx           # Home screen (camera/gallery + analyze)
  results.tsx         # Recipe results screen
src/
  components/         # Reusable UI components
  hooks/              # TanStack Query hooks
  lib/                # API clients (OpenAI, Supabase, React Query)
  services/           # Business logic (AI calls, storage)
  types/              # TypeScript types and Zod schemas

How It Works

1. Capture - Take a photo of your fridge or pick one from your gallery
2. Analyze - The image is sent to GPT-4o's vision API, which identifies all visible food items
3. Generate - The identified ingredients are sent back to GPT-4o, which generates 3-5 recipes
4. Browse - View recipe cards with difficulty ratings, calorie counts, ingredient lists, and cooking steps