Inspiration
We wanted to solve two problems: food waste and the difficulty of deciding what to cook. We realized that many people have perfectly good ingredients sitting in their fridge but lack the inspiration to combine them into a healthy meal.
What it does
LeftoverHero takes a list of ingredients you already have and uses a local LLM (Ollama) to generate a customized, healthy recipe. It provides prep time, health benefits, and a step-by-step guide.
How we built it
We built a lightweight web frontend using vanilla HTML/JS for speed and simplicity. The backend logic runs entirely on the user's machine using Ollama, ensuring privacy and zero cloud costs. We used llama3 for its reasoning capabilities to create culinary creative recipes.
Challenges we ran into
Connecting a web app to a local AI API presented CORS and mixed-content challenges. We initially tried tunneling but realized a direct localhost connection was faster and more private, so we re-architected the app to be "Local-First".
Accomplishments that we're proud of
We successfully integrated a powerful LLM into a simple web interface that anyone can use to reduce food waste.
What's next for LeftoverHero
We plan to add image recognition so users can just snap a photo of their fridge content!
Built With
- css3
- html5
- javascript
- llama3
- ollama

Log in or sign up for Devpost to join the conversation.