Skip to content

README.md

Latest commit

 

History

History
83 lines (58 loc) · 3.98 KB

README.md

File metadata and controls

83 lines (58 loc) · 3.98 KB

🏟️ CrowdFlow AI: Intelligent Stadium Management

Deployment Node.js Express Docker License

CrowdFlow AI is a professional-grade full-stack solution designed to revolutionize the way large-scale sporting venues manage crowd dynamics. By combining real-time data simulation with a context-aware AI assistant, it ensures safer, smoother, and more efficient event operations.

🎯 Chosen Vertical: Smart Stadiums & Event Management

This project focuses on the Sports & Large-scale Events vertical. Managing tens of thousands of people in real-time requires more than just static monitoring—it requires dynamic, actionable intelligence. CrowdFlow AI addresses the specific pain points of venue operators:

  • Safety: Identifying and mitigating critical bottleneck zones.
  • Experience: Reducing wait times at concessions and facilities.
  • Efficiency: Optimizing visitor flow through intelligent routing.

🧠 Approach and Logic

The core logic of CrowdFlow AI is centered around Dynamic Context Integration.

  1. Simulated State Management: The backend maintains a live "world state" of the stadium, updating occupancy, wait times, and congestion levels every 2.8 seconds.
  2. Context-Aware Parsing: The "Ask Me" agent uses a specialized natural language processor (NLP) logic that maps user intent directly to the current state of the stadium data.
  3. Proactive Heuristics: If a user asks about a facility with a high wait time, the system doesn't just report the time—it proactively searches for and suggests a better alternative.

⚙️ How the Solution Works

CrowdFlow AI is architected as a high-performance monolith for low-latency synchronization between data and visualization.

1. The Real-Time Pipeline

graph LR
    A[Data Simulation] -->|Update State| B(Express Server)
    B -->|API Response| C[Frontend Dashboard]
    C -->|User Interaction| B
    B -->|AI Processing| D[Chat Response]
Loading

2. Frontend Layer

Built with Vanilla JS for maximum performance, the frontend polls the backend every 2.8 seconds. It performs direct DOM manipulation to update the SVG heatmap and statistical cards without full-page reloads, ensuring a "live" feel.

3. Backend Layer (Monolith)

The Node.js/Express backend serves as both the API provider and the static file host. This monolithic approach ensures that the UI and API are always in sync and simplifies the deployment to a single containerized service.

4. Cloud Infrastructure

Containerized via Docker and deployed to Google Cloud Run, the application scales automatically to handle event-day traffic spikes.

📋 Any Assumptions Made

During development, the following operational assumptions were made:

  • Simulated Cadence: A 2.8-second update frequency is assumed to be an optimal balance between "real-time" responsiveness and network efficiency for stadium monitoring.
  • Keyword Entry Points: It is assumed that users will use primary identifiers (e.g., "North Stand", "Food Court", "Gate E") when interacting with the AI agent.
  • Global State: The simulation assumes a unified state for all dashboard users, simulating a central command center environment.

📦 Installation & Deployment

Local Setup

git clone https://github.com/ShreyashChaugule-github/CrowdAI.git
cd CrowdAI/backend
npm install
npm run dev

Cloud Run Deployment

gcloud run deploy crowdflow-ai --source . --project promptwars-493509

📄 License

This project is licensed under the MIT License.