← Back to Projects

SafeBelly

AI/MLFull StackBackend

Checking whether a product is safe during pregnancy means decoding ingredient labels by hand.

Reading ingredient labels while pregnant is guesswork. Built a scanner that gives a trimester aware verdict with the reasoning, and won the Cursor Toronto Hackathon with it. Scan a barcode, photograph a label, or paste ingredients and get a verdict for your trimester. Every ingredient is checked against a hand reviewed list first, and only unknown ones go to the AI, which is told to answer unknown instead of guessing. Built with Django REST Framework and React, with barcode lookups across three public product databases and Claude and Gemini reading label photos.

SafeBelly screenshot 1

Tech Stack

Django REST FrameworkReactClaude APIGemini APIVite

Key Features

Barcode, Photo and Text Scan
Trimester Aware Verdicts
Curated Ingredient List
Honest Unknown Results