A multimodal full-stack web application that leverages AI to create and customize recipes. It allows users to generate personalized recipes by either describing their needs in natural language or by uploading a picture of their ingredients. The project features a sleek, user-friendly interface with modern and classic layouts to provide a seamless user experience across different devices.

• Websitehttps://kaifan.vercel.app/
• Features
  • Dual Input Modes: The application offers two primary ways to generate a recipe: a text-based chat interface and an image recognition tab.
  • AI-Powered Ingredient Recognition: Users can upload a photo of their ingredients, and the system automatically identifies them using a vision model, offering a list of identified ingredients that can be edited.
  • Conversational Recipe Optimization: The app acts as an "AI chef," allowing users to refine and optimize a generated recipe by chatting with the system. It can handle requests such as "simplify the steps," "reduce calories," or "add more vegetables".
  • AI Image Generation: Once a recipe is finalized, the application can generate a high-quality visual representation of the dish, creating a more engaging user experience.
• Technology
  • Frontend: Next.js, React, Tailwind CSS, TypeScript.
  • Backend: FastAPI, Rest API.
  • Models: GPT-4o-mini, Qwen-image.