The Problem: Food waste is a huge issue. People often have random ingredients but don't know how to combine them, leading to wasted food and money.
The Solution: An AI-powered kitchen assistant that uses the Gemini 2.5 Flash model to instantly generate structured, creative recipes based on whatever the user has on hand.
How I built it:
Language: Python
Model: Google Gemini 2.5 Flash (chosen for its speed and reasoning).
Challenges I ran into: "One major challenge was navigating API Versioning. I initially encountered a 404 error when trying to access older models. I overcame this by writing a diagnostic script to query the Gemini API's supported actions, which helped me identify that my account had access to the cutting-edge Gemini 2.5 and 3.0 models. I also learned the importance of Environment Variable security after realizing how easy it is to accidentally expose API keys in source code."
Security: Implemented .env environment variables to protect API credentials.
Logic: Used advanced Prompt Engineering to ensure structured Markdown output.
What I learned: I learned how to handle API errors, manage hidden environment variables, and how to "steer" an AI model using specific formatting instructions.
Log in or sign up for Devpost to join the conversation.