Generative Recipe Prototype
Dishire — Context & Emotion-Aware
AI Recipe Generation Prototype
Your Emotion. Your Story. Your Dish.
Dishire is an AI-powered Food Decision Maker that helps decide “what to eat right now” by considering a user’s context, preferences, basic profile, and constraints.
In this prototype, I implemented GPT-based prompt templates and a multi-step recipe generation pipeline, delivering an end-to-end flow that recommends recipes for different meal situations.
Multi-step LLM Generation
Implemented a pipeline that generates menu direction, ingredients, and step-by-step instructions in separated stages for more controllable outputs.
Implemented a pipeline that generates menu direction, ingredients, and step-by-step instructions in separated stages for more controllable outputs.
Context-based Prompt Routing (Prototype)
Users select a prompt type (e.g., standard recommendation, ingredient-based), and the system generates a recipe aligned with the selected situation.
Users select a prompt type (e.g., standard recommendation, ingredient-based), and the system generates a recipe aligned with the selected situation.
Emotion & Profile-aware Personalization (Vision)
Designed with a longer-term goal of incorporating emotion, health goals, and dietary constraints into a personalization engine (future direction).
Designed with a longer-term goal of incorporating emotion, health goals, and dietary constraints into a personalization engine (future direction).
Prototype UI
Dishire prototype main screen — entry point for starting recipe generation.