VenueFlow AI

🏟️ I built VenueFlow AI — real-time crowd intelligence for 100,000+ capacity stadiums.

Built for the Google Build with AI program. Here's what's under the hood:

⚡ The Core Problem
How do you route 105,000 fans to the right gate — instantly, without chaos?

🧠 The Solution: Min-Heap Priority Queue
Instead of brute-forcing every gate option, I implemented a composite scoring algorithm:

score = queue_length / throughput + α · distance

→ O(N + k log N) time complexity
→ Each routing query: ~O(log 20) ≈ constant time
→ Full rebalance of 105K users: ~450,000 ops in < 50ms

That's not a guess — that's math.

🏗️ Full-Stack, Production-Grade
• Python Flask + WebSockets backend
• React 19 + Tailwind CSS 4 frontend
• Google Gemini AI for the fan assistant & crowd context
• Dual persistence: Redis → falls back to local JSON automatically
• Google Maps API with Haversine fallback for zero-downtime routing

🤖 AI-Powered
Every chat query injects live crowd context into Gemini — so fans get answers that actually reflect what's happening in the venue right now.

📍 The UX Insight
I used cardinal directions (N/S/E/W) as spatial anchors — because fans already think this way inside a circular stadium. Reduced time-to-decision without a single extra click.

Building this taught me that the right data structure isn't a detail — it's the whole product.

link : https://venueflow-service-1015826998904.us-central1.run.app/