V-Fridge: Smart Fridge Ingredient Management and Recipe Recommendation System Based on Gemini AI

V-Fridge is an AI-driven smart fridge management system integrated with Google Gemini AI, built using Next.js15. It addresses common kitchen pain points like food waste and lack of cooking inspiration by offering features such as ingredient inventory tracking, expiration reminders, AI-generated recipes, Redis caching for performance, and Zod validation for security. Key tech stack includes Next.js15, Gemini AI, PostgreSQL, Drizzle ORM, Redis, etc.

In modern fast-paced life, family ingredient management often becomes an overlooked pain point: food forgotten until expiration, not knowing what to cook with available ingredients, repeated purchases leading to waste. These issues are common in many households, and V-Fridge aims to address these daily problems with an intelligent solution.

1. AI Chef Assistant: Powered by Gemini AI, it generates creative recipes based on existing ingredients, considering expiration priority, nutrition balance, and taste preferences (e.g., prioritizing milk near expiration).
2. Smart Inventory Management: Users can add ingredients with quantity, unit, and expiration date; the system tracks remaining shelf life and reminds users to consume priority items.
3. Privacy-Focused History: Chat records auto-delete after 24h to protect user privacy (dietary preferences, allergies) and reduce performance burden.
4. Performance Cache: Upstash Redis is used for fast chat history access and API rate limiting, ensuring smooth user experience.

• Frontend: Next.js15 with App Router (server-side rendering, parallel data fetching, streaming).
• State Management: Zustand (lightweight, no complex config).
• Style: Tailwind CSS + Shadcn UI (accessible components).
• Data Validation: Zod (client & server-side, end-to-end type safety).
• Database: PostgreSQL (Neon DB) + Drizzle ORM (type-safe operations).
• Auth: NextAuth.js (multi-login support, session management).
• AI Integration: Google Gemini SDK (text generation API).
• Cache/Limiting: Upstash Redis (session cache, API rate limiting).

1. Type Safety: Entire codebase uses TypeScript, from DB schema to API responses and UI props, catching errors early.
2. Layered Validation: Zod schema for client-side form validation (instant feedback) and server-side revalidation (security).
3. API Protection: Upstash middleware for Gemini API rate limiting to prevent abuse and control costs.

• Family Users: Helps manage daily ingredients, reduce waste, get cooking inspiration (ideal for busy office workers).
• Shared Accommodation: Roommates can co-manage public fridge, track individual ingredients, avoid confusion/waste.
• Small Food Businesses: Lightweight inventory management for small restaurants/cafes, plus AI-recommended daily specials.

V-Fridge demonstrates how AI large models integrate with traditional apps—Gemini is not just a chatbot but a core smart engine. Its tech choices (Next.js15 App Router, Zustand, Drizzle ORM) reflect modern development trends. For developers, it's an excellent reference for learning full-stack development, showing how to combine tech to solve real problems. Future directions may include deeper AI integration and expanded use cases.