Burning Events - Project Overview

Inspiration

San Francisco's tech scene is full of events, but finding relevant ones is hard. People often miss great conferences or meetups because they don't know where to look or can't describe exactly what they want. We wanted to make event discovery as simple as asking "What AI events are happening in SoMa this weekend?"

What it does

Burning Events is an AI-powered event discovery platform for San Francisco. Users can search for tech events using natural language queries like "networking events for startups" or browse an interactive map showing events across SF neighborhoods. The platform combines intelligent search with precise location filtering to help people find conferences, workshops, meetups, and social events.

How we built it

We built it using the T3 Stack (Next.js, TypeScript, tRPC, Prisma) with a PostgreSQL database using pgvector for semantic search. The AI search uses OpenAI's GPT to extract keywords from natural language queries, then searches event descriptions using vector embeddings. The interactive map uses Leaflet to display events with neighborhood-level filtering across San Francisco areas.

Challenges we ran into

The biggest challenge was creating accurate semantic search that understands event context. Balancing AI keyword extraction with traditional filtering was tricky - we needed the AI to be smart but not override user preferences. We also had to handle San Francisco's complex neighborhood geography and ensure the map interactions felt smooth while maintaining search performance. We also spent time on providing a "best match".

Accomplishments that we're proud of

We successfully combined AI-powered search with geographic filtering in a way that feels natural. The semantic search accurately finds relevant events even when users don't use exact keywords. The neighborhood-based filtering system captures San Francisco's unique geography, and the real-time search with map visualization creates an intuitive discovery experience.

What we learned

We learned how to effectively combine vector embeddings with traditional database queries for hybrid search. Working with geographic data taught us about balancing precision with usability. We also discovered the importance of fallback mechanisms when AI services fail, and how to design prompts that extract useful keywords without losing user intent.

What's next for Burning Events

We plan to add user accounts with personalized recommendations, event calendar integration, and real-time notifications for new events matching user interests. We also want to expand beyond San Francisco to other tech hubs, add event organizer tools for posting events, and implement social features like attendee networking and event reviews.

Built With

• maps
• postgresql
• t3
• vectors