MiamiTechConnect: The AI-Powered Tech Event Discovery Platform

Inspiration

Miami’s vibrant tech scene is growing rapidly, but finding relevant events can be overwhelming. We were inspired to create MiamiTechConnect to simplify event discovery, leveraging AI to match users with events that align with their interests. Our goal was to build a tool that not only lists events but personalizes the experience, making it easier for tech enthusiasts, professionals, and newcomers to connect with Miami’s tech community.

What it does

MiamiTechConnect is a web application that helps users explore upcoming tech events in Miami. It offers:

• Personalized Curation: Users set interest keywords (e.g., "AI", "Women in Tech") to see tailored event recommendations under "Curated Events."
• Smart Search: A semantic search engine finds events based on natural language queries (e.g., "blockchain meetups"), ranked by relevance.
• Comprehensive Listings: The "All Events" tab displays every upcoming event chronologically.
• Rich Details: Each event includes an image, start time with duration, price, location, and a direct RSVP link.

The platform combines AI-driven search with a visually appealing, user-friendly interface to enhance event discovery.

How we built it

• Backend: We used Flask to create a lightweight API with two endpoints:
  • /search: Employs FAISS and Sentence Transformers (all-MiniLM-L6-v2) for semantic similarity search over event data.
  • /all_events: Returns all events from a preprocessed JSON file.
• Data Processing: embedder.py transforms raw event data (Upcoming_tech_events_Miami.json) into embeddings, indexed with FAISS, and saves it as events_with_ids.json.
• Frontend: Built with HTML, Tailwind CSS, and JavaScript:
  • A responsive UI displays event cards with images, times, prices, and RSVP links.
  • An "Interests" modal lets users set keywords, stored in localStorage, driving curated recommendations.
• Workflow: We processed a JSON dataset of Miami tech events, integrated AI search, and designed an interactive frontend, all within the hackathon timeframe.

Challenges we ran into

• Time Parsing: Integrating start_time and end_time with proper formatting and duration calculation required debugging to avoid undefined errors.
• Scoring Accuracy: Tuning FAISS distances and frontend scoring to rank events accurately based on user queries and interests was tricky, requiring multiple iterations.
• UI Design: Balancing rich event details (images, times, prices) in a clean layout pushed us to refine Tailwind CSS usage under time constraints.
• Dynamic Curation: Replacing static samples with an interests-based system meant rethinking data flow and ensuring seamless API integration.

Accomplishments that we're proud of

• AI-Powered Personalization: Successfully implemented a keyword-based curation system that adapts to user preferences using FAISS.
• Polished UI: Delivered a visually appealing interface with event images and detailed info, enhancing user engagement.
• Robust Search: Achieved reliable semantic search with relevance scoring, surpassing basic keyword matching.
• Hackathon Speed: Built a fully functional prototype—from data processing to frontend—in a short timeframe, showcasing teamwork and adaptability.

What we learned

• AI Integration: Gained hands-on experience with FAISS and Sentence Transformers for semantic search, deepening our understanding of vector-based similarity.
• Frontend Dynamics: Mastered dynamic UI updates with JavaScript and localStorage for persistent user data.
• Time Management: Learned to prioritize features (e.g., curation over event submission) under tight deadlines.
• Debugging: Improved skills in tracing issues across backend-frontend interactions, like fixing start_time rendering.

What's next for The AI-Powered Tech Event Discovery Platform

• Event Submission: Add a form for users to submit new events, expanding the database dynamically.
• User Accounts: Implement authentication to save interests and preferences persistently across devices.
• Enhanced Filters: Introduce filters for price, date range, and categories to refine search and curation.
• Mobile App: Develop a mobile version for on-the-go event discovery.
• Real-Time Updates: Integrate a database or API for live event data instead of static JSON.

Project Structure

Built With

• faiss
• fastapi
• huggingface
• python
• tailwind