Food-Culture Map

Explore the cultural journeys of food through AI-powered stories and an interactive world map

Comment

🌍 Inspiration

We aimed to combine semantic AI with vector search in MongoDB to unearth food stories that are not just historically accurate but contextually relevant to the user’s query.

🤖 What it Does

Using MongoDB’s cutting-edge vector search, it retrieves the most semantically similar food narratives based on the user’s question, ensuring highly relevant and culturally rich results.

🛠️ How We Built It

Architecture Diagram

Frontend:

  • React.js + TypeScript for a dynamic UI.
  • Integrated Mapbox for animated navigation and smooth map interactions.

Backend:

  • Express.js (Node.js) handles API routing, user queries, and MongoDB interactions.
  • FastAPI (Python) manages embedding generation using TensorFlow’s Universal Sentence Encoder.

Database:

🔥 MongoDB with Vector Search – the core engine behind our semantic search.

  • We stored high-dimensional embeddings (from TensorFlow USE) alongside food documents.
  • These embeddings enable semantic retrieval — not just keyword matching, but true contextual similarity.

This formed the backbone of our Retrieval-Augmented Generation (RAG) pipeline, seamlessly connecting raw data to cultural storytelling.

AI Integration:

  • Used Google's Gemini API to generate rich, human-like narratives from semantically retrieved data.

Deployment:

  • Fully containerized using Docker.
  • Hosted on Google Cloud Platform.

⚔️ Challenges We Ran Into

  • Ensuring accuracy of vector search and relevance of results using embeddings.
  • Designing an intuitive yet informative UI that combines storytelling and interactive maps.

🏆 Accomplishments That We're Proud Of

  • Successfully implemented an end-to-end RAG pipeline with MongoDB, TensorFlow, and Gemini.
  • Delivered a visually rich and user-friendly experience.
  • Enabled semantic query understanding using real-time embeddings and vector search.

📚 What We Learned

  • How to leverage semantic embeddings and vector similarity search in MongoDB.
  • Real-time LLM integration with production-ready pipelines.
  • The power of combining maps, AI, and visual storytelling to preserve and present cultural heritage.

🚀 What’s Next for Food-Culture Map

  • Order Food Using Swiggy API: Allow users to ORDER the food shown in the story, from swiggy/zomato.
  • Multilingual Support: Generate narratives in native languages for better global reach.
  • Gamification: Add quizzes and interactive timelines to boost user engagement.
  • Mobile App: Launch a mobile-friendly version with offline story viewing.
  • AR Integration: Explore augmented reality food journeys for use in education and tourism.

Built With

Share this project:

Updates