DEV Community

Sai Jadhav
Sai Jadhav

Posted on

🛡️ SafeBite: Building an AI-Powered Cloud Allergy Scanner for a Loved One

#hacktoberfest #python #devchallenge #hf26challenge

Challenge Entry: Hacktoberfest Weekend Challenge — Build for a Friend 🤝

Live App: [https://safebite-jrr6.onrender.com/]

GitHub Repository: https://github.com/12jadhavsai2006/safebite

💡 The Inspiration: Why I Built This

Grocery shopping with severe food allergies or strict dietary restrictions can be genuinely stressful. Reading through endless, tiny ingredient lists on product packaging while trying to spot hidden derivatives or cross-contamination warnings is exhausting.

I built SafeBite for someone close to me who deals with strict dietary restrictions. I wanted to give them a tool they could open instantly on their phone in the middle of a grocery aisle, snap a photo of a label, and get an immediate, crystal-clear safety verdict without second-guessing.

🛠️ How It Works & Why Open-Source AI Matters

SafeBite relies on open innovation to make fast, reliable inference possible:

  • The AI Core: Powered by open-weight vision models accessed via the Groq API (qwen/qwen3.8-27b), allowing lightning-fast multimodal analysis of ingredient images.
  • Frontend & UX: Built entirely in Python using Streamlit for a clean, mobile-optimized dashboard.
  • Cloud Infrastructure: Hosted on Render, making the live app accessible globally on any mobile browser with zero setup required for the end user.

Using an open-weight model approach meant I could precisely prompt the model for strict safety checks (SAFE vs DANGER verdicts) while keeping the architecture lightweight, flexible, and completely transparent.

✨ Key Features

  • Mobile Camera Integration: Users can snap live photos directly from their phone camera in the store.
  • Preset & Custom Profiles: Pre-loaded options for severe allergies, gluten restrictions, and vegan diets, alongside custom input fields.
  • Recent Scans History: A local session log to easily reference previous product checks during a shopping trip.

🚀 Check It Out

The project is 100% open-source and live right now. Feel free to check out the repository or test the live app. Feedback and pull requests are always welcome!

Happy hacking, everyone! 🛡️✨

Top comments (0)