AI Web Application · 2025

Kulkas Berisi

Zero-Waste AI web application that transforms leftover fridge ingredients into delicious recipes before expiry.

Context

Many households and students discard leftover fridge ingredients simply because they don't know what dish to cook with them.

Kulkas Berisi was built as a smart Zero-Waste AI platform recommending step-by-step recipes tailored directly to available fridge items.

Household food waste reaches tons every month due to a lack of quick and practical recipe ideas.

Traditional recipe apps require manual search and rarely accommodate limited or random ingredient combinations.

Engineered an AI-powered recipe generator that takes a list of fridge ingredients and instantly outputs structured step-by-step cooking instructions.

Designed a modern, intuitive, and fully responsive landing page & app interface with zero-waste impact metrics.

Built the full application architecture from UI/UX design, AI prompt engineering, to cloud edge deployment.

Implemented real-time API integrations for recipe generation and usage statistics.

Leveraged Next.js Server Actions to securely process AI prompts on the server without exposing API keys to the client.

Enforced strict JSON schema output formatting on LLM responses to ensure reliable client-side parsing.

Crafted an aesthetic light-theme landing page with interactive preview cards demonstrating recipe generation (e.g., Egg - Chicken - Carrot).

Initial LLM response times took several seconds. I implemented loading states and popularity caching for near-instant responses.