Inspiration

Many people struggle to quickly determine if a food item is safe due to allergies, dietary restrictions, or health goals. We wanted to create an instant, reliable, and accessible solution that turns any ingredient label into personalized insights.

What it does

NutraSpec lets users snap a photo of a food label and instantly get a personalized safety analysis. It highlights allergens, scores ingredients from low to high risk, provides a human-readable summary, and even reads results aloud for accessibility.

How we built it

Frontend: React 19, TypeScript, Vite, Tailwind CSS v4 Backend: Python 3.11+, FastAPI, Pydantic, Uvicorn AI / Vision: Google Gemini for text extraction and ingredient analysis Voice: ElevenLabs text-to-speech Database: Supabase (user profiles + analysis cache) We combined AI-powered analysis with deterministic allergen checks to ensure critical safety.

Challenges we ran into

Extracting ingredients accurately from diverse label formats Merging AI analysis with deterministic allergen safety checks Making TTS summaries clear and natural-sounding

Accomplishments that we're proud of

Full AI + deterministic safety pipeline built in one hackathon Accessible experience with camera scanning + spoken results Seamless integration of multiple APIs into a single workflow

What we learned

Real-world ingredient data is messy and inconsistent Combining AI with deterministic rules improves reliability Accessibility features like TTS are valuable for usability

What's next for NutraSpec

Improve AI risk scoring granularity and recommendations Expand personalized nutrition insights and dietary suggestions

Built With

  • elevenlabs(tts)
  • fastapi
  • frontend:-react-19
  • gemini
  • postgresql
  • pydantic
  • pydantic-ai-/-vision:-google-gemini-(vision-+-ingredient-analysis)-voice:-elevenlabs-tts-database:-supabase-(postgresql
  • python
  • supabase
  • tailwind-css-v4-backend:-python-3.11+
  • typescript
  • uvicorn
  • vite
Share this project:

Updates