RecipeRAG — Multimodal AI Recipe Creator & Personal Chef Agent
Client: MetaLog (Product Project)
The Challenge
Home cooks waste ingredients and struggle to invent recipes from what they already have. Existing recipe sites are static — they cannot see what is in your kitchen, respect your allergies, or adapt to your dietary preferences in real time.
Our Solution
A Next.js 16 web app powered by Gemini 3 that analyzes ingredient photos captured via device camera and generates personalized recipes instantly. RAG-grounded suggestions from real recipes, voice-to-text input, persistent chat, and dietary restriction/allergy management. FastAPI backend with Neon PostgreSQL + pgvector, deployed on Vercel and Railway, plus a social platform for sharing AI-enhanced recipes.
The Result
A complete personal chef agent that reduces food waste by using available ingredients, adapts every recipe to user needs, and turns cooking into an accessible, sustainable, enjoyable experience for all skill levels.
Gemini 3
AI Model
RAG + pgvector
Retrieval
Camera + Voice
Input Modes
Vercel + Railway
Deploy
