Inspiration
San Francisco’s Chinatown is rich in history, culture, and incredible food, but small businesses often struggle with visibility in the digital age. Many tourists and even locals rely on word of mouth or big-name review sites, which often push traffic toward mainstream, tourist-heavy spots. We wanted to create a solution that highlights hidden gems while supporting authentic, family-owned businesses.
What it does
Chinatown Catalyst is an AI-powered restaurant recommender designed to connect people with the best, most affordable, and authentic dining experiences in Chinatown. Users can input a query like "cheapest dumplings" or "best dim sum under $15", and the system instantly provides personalized recommendations along with Google Maps links for easy navigation.
How we built it
Data Collection: We scraped restaurant details, menus, and pricing data. AI Model: We used SentenceTransformers to generate text embeddings. Search Engine: FAISS was implemented for fast, similarity-based recommendations. Backend: Built with FastAPI to handle queries. Frontend: A Streamlit-powered web app for an intuitive user experience.
Challenges we ran into
Data gaps: Some restaurants lacked online menus, requiring manual data curation. AI fine-tuning: Generating precise, relevant recommendations from unstructured data. Integration issues: Connecting AI embeddings, FAISS search, and API requests seamlessly.
Accomplishments that we're proud of
Successfully deployed a working AI-powered recommender in a short time. Created a fast, intuitive interface that makes finding great Chinatown eats easy. Built something that directly supports small businesses and boosts local economies.
What we learned
Data quality matters—good recommendations rely on clean, well-structured data. Optimizing AI embeddings significantly improves search speed and relevance. Streamlining APIs can make AI-powered applications scalable and efficient.
What's next for ChinaTown Catalyst
Live web deployment to make the tool available for public use. Expanding restaurant coverage with real-time menu updates. User reviews & engagement to further improve recommendations. Business partnerships to provide exclusive discounts & offers through the platform.
Built With
- anthropic
- python
- rag
Log in or sign up for Devpost to join the conversation.