ingreBOARD

Context

Consumer packaged foods use complicated chemical names and layout variations on ingredient labels, making it hard to identify allergens and additives.

Motivation

I wanted to build a fast, privacy-first mobile scanner that decodes ingredient lists using OCR and LLMs in sub-second times.

Outcome

An AI-powered scanner running Llama 3.3 via Groq on Vercel Edge Functions, combining hybrid OCR, barcode integration, and zero-retention ephemeral privacy.