Skip to content

snarayan-redhat/CCS-AI-Lab

BranchesTags
 
 

Repository files navigation

SkyRescue AI Lab - Software Department

"First to the scene, first to save"

Company: SkyRescue - AI-First Emergency Response Drone Startup Product: Cardiac arrest response drones Mission: Deliver AEDs to cardiac arrest victims in <3 minutes via autonomous drones with voice AI guidance

🚁 Company Overview

SkyRescue deploys AI-powered drones that respond to 911 cardiac arrest calls:

  • Speed: 2-3 minute response vs 8-12 minute ambulance ETA
  • Innovation: Voice AI talks bystanders through AED deployment and CPR
  • Impact: Potential to save 50,000-75,000 additional lives per year in the US

Use Case: When someone calls 911 for cardiac arrest, a dispatcher launches a drone that flies autonomously to the location, delivers an AED, and the onboard voice AI guides the bystander through life-saving procedures while waiting for the ambulance.

💙 Core Company Values

Value Principle
AI-Powered Velocity Speed over Perfection - Use AI to create working prototypes in hours, not months
Uncompromising Safety Fail Safe, Not Fast - Multiple redundant safeguards for life-critical operations
Human-AI Partnership Augment, Don't Replace - AI amplifies human creativity and judgment
Radical Transparency Documentation is Our Default - Build high-quality data foundation for AI systems

📁 Software Department Deliverables

This repository contains all technical materials developed by the Software Department for the AI Lab Workshop:

✅ Completed Work

  1. UI Mockups (mockups/)

    • Dispatcher console with SkyRescue color scheme
    • Live voice AI transcript interface
    • Real-time telemetry dashboard
    • Emergency assessment and launch controls

  2. API Documentation (api-docs/)

    • Complete OpenAPI 3.0 specification
    • MQTT integration guide with examples
    • Real-time WebSocket endpoints
    • 911 CAD system integration

  3. Interactive Prototype (prototype/)

    • Fully functional HTML/CSS/JS dispatcher console
    • Simulated mission from launch to AED deployment
    • Live voice AI conversation demonstration
    • Interactive controls (mute, override)

  4. Technical Architecture

    • Edge AI stack (NVIDIA Jetson Orin Nano)
    • Voice AI system (Whisper + Phi-3 + Piper)
    • Cloud backend (AWS IoT Core + Lambda)
    • Hybrid edge/cloud architecture

  5. Presentation Materials

    • Complete workshop presentation guide
    • Technical deep-dive slides
    • Demo script and talking points
    • Q&A preparation

🗂️ Directory Structure

ai_lab/
├── mockups/                          # UI/UX Design (D2 + PlantUML)
│   ├── dispatcher-console.d2         # Main dispatcher interface
│   ├── dispatcher-console.svg        # Generated mockup
│   ├── drone-dashboard.d2            # Original drone control UI
│   ├── mission-planning.d2           # Mission planner interface
│   └── dashboard.salt                # Detailed wireframe
│
├── api-docs/                         # API Documentation
│   ├── skyrescue-api.yaml            # OpenAPI 3.0 specification
│   ├── MQTT_INTEGRATION.md           # Real-time drone communication
│   └── README.md                     # API quick start guide
│
├── prototype/                        # Interactive Prototype
│   ├── dispatcher-console.html       # Full working demo
│   └── README.md                     # Usage instructions
│
├── Technical Documentation           # AI Architecture
│   ├── SKYRESCUE_AI_ARCHITECTURE.md  # Edge + Cloud AI design
│   ├── DISPATCHER_WORKFLOW.md        # 911 dispatcher integration
│   └── VOICE_AI_SYSTEM.md            # Voice AI technical specs
│
├── PRESENTATION.md                   # Workshop presentation guide
├── generate-mockups.sh               # Build script for UI mockups
└── README.md                         # This file

🚀 Quick Start

1. Generate UI Mockups

cd ~/ai_lab
./generate-mockups.sh

This creates SVG and PNG files from declarative D2/PlantUML sources.

2. View Interactive Prototype

cd ~/ai_lab/prototype
python3 -m http.server 8000
# Open http://localhost:8000/dispatcher-console.html

Click "LAUNCH SKYRESCUE" to see the complete mission simulation:

  • 2:15 minute autonomous flight
  • Real-time telemetry updates
  • Voice AI conversation with bystander
  • AED deployment and guidance

3. Explore API Documentation

# View in Swagger UI (online)
# Go to https://editor.swagger.io/
# Import api-docs/skyrescue-api.yaml

# Or run mock API server
npm install -g @stoplight/prism-cli
prism mock api-docs/skyrescue-api.yaml

🎨 Color Scheme

SkyRescue uses a Medical + Technology palette:

  • Primary: Sky Blue (#00AEEF) - Trust and calmness
  • Secondary: White (#FFFFFF) - Clean and clear
  • Accent: Neon Green (#39FF14) - Futuristic and attention-grabbing
  • Dark: Charcoal (#333333) - Text and backgrounds

🤖 AI Technology Stack

Edge AI (On-Drone)

Hardware: NVIDIA Jetson Orin Nano (8GB) - $499

  • Computer Vision:
    • YOLOv8n (person detection) - <20ms
    • DeepLabV3 (landing zones) - <30ms
    • Stereo vision (obstacle avoidance) - <50ms
  • Voice AI:
    • Whisper-tiny (speech-to-text) - 95% accuracy, 99 languages
    • Hybrid FSM + Phi-3-mini (dialogue) - 100-200ms
    • Piper TTS (text-to-speech) - <50ms
  • Total: ~10W power, <250ms latency, 100% offline capable

Cloud AI (Backend)

Platform: AWS (HIPAA-compliant)

  • Drone selection and routing
  • Risk assessment (XGBoost)
  • Fleet coordination
  • 911 CAD system integration
  • Model training and OTA updates

Cost: ~$3/drone/month

📊 Key Metrics

Business Impact

  • US Cardiac Arrests/Year: ~350,000
  • Out-of-Hospital: 245,000 (70%)
  • Current Survival Rate: ~10% (24,500)
  • With SkyRescue: +20-30% survival increase
  • Additional Lives Saved: 49,000-73,500 per year

Technical Performance

  • Drone ETA: 2-3 minutes
  • Ambulance ETA: 8-12 minutes
  • Time Advantage: 5-6 minutes faster
  • Voice AI Latency: <250ms (conversational)
  • Landing Zone Confidence: 92%+
  • Multi-language Support: 10+ languages auto-detected

Cost Analysis

  • Drone Hardware: $12,000 (one-time)
  • Voice AI Hardware: $140/drone (mic array + speaker)
  • Cloud Services: $3/drone/month
  • Cost per Rescue: $20-30 (operational)

🎯 Workshop Demonstration

Demo Flow (15 minutes)

Phase 1: Emergency Call (2 min)

  • 911 dispatcher receives cardiac arrest call
  • AI recommendation appears: LAUNCH (95% confidence)
  • Show ETA comparison: Drone 2:15 vs Ambulance 8:00

Phase 2: Launch (1 min)

  • Dispatcher clicks "LAUNCH SKYRESCUE"
  • Drone acknowledges and takes off
  • Live telemetry starts streaming

Phase 3: Autonomous Flight (2 min)

  • Real-time position updates
  • Altitude: 0→120m
  • Speed: 15 m/s
  • AI detections: obstacles, people
  • Landing zone detection: 92% confidence

Phase 4: Voice AI Activation (5 min)

  • Drone arrives and lands
  • Voice AI: "This is SkyRescue. I have an AED. Can you help?"
  • Bystander: "Yes! What do I do?!"
  • Guides through: breathing check → AED retrieval → activation
  • AED device takes over with shock/CPR instructions

Phase 5: Interactive Features (2 min)

  • Show mute/unmute control
  • Demonstrate manual voice override
  • Highlight live transcript streaming

Key Selling Points:

  • 🚀 Speed: 5:45 faster than ambulance
  • 🗣️ Voice AI: Guides untrained bystanders
  • 🌍 Multi-language: Auto-detects 99 languages
  • 📡 Offline: Works without cell signal
  • 🏥 Safety: Pre-scripted for medical accuracy

📖 Documentation

For Technical Deep-Dive

  • SKYRESCUE_AI_ARCHITECTURE.md - Complete AI stack details
  • DISPATCHER_WORKFLOW.md - 911 integration workflow
  • VOICE_AI_SYSTEM.md - Voice AI implementation
  • api-docs/MQTT_INTEGRATION.md - Real-time communication

For API Development

  • api-docs/skyrescue-api.yaml - OpenAPI specification
  • api-docs/README.md - API quick start guide

For Prototype Customization

  • prototype/README.md - Interactive demo usage
  • mockups/ - Declarative UI source files

For Presentation

  • PRESENTATION.md - Complete workshop script with Q&A

🛠️ Tools & Technologies

UI/UX Design

  • D2 - Declarative diagrams (modern, easy)
  • PlantUML Salt - Detailed wireframes
  • Install: See generate-mockups.sh

API Development

  • OpenAPI 3.0 - REST API specification
  • MQTT - Real-time drone communication
  • AWS IoT Core - Device management

Prototype

  • HTML5/CSS3/JavaScript - Pure vanilla, no frameworks
  • WebSocket - Real-time updates
  • Single file - Easy deployment

AI/ML Stack

  • TensorRT - Inference optimization
  • Whisper - Speech recognition
  • Phi-3 - Small language model
  • Piper - Text-to-speech

🎤 Presentation Tips

Key Messages

  1. Time saves lives: 5:45 faster delivery = 20-30% more survivors
  2. Voice AI is the differentiator: No other drone talks bystanders through survival
  3. Edge-first = Reliability: Works offline, <250ms latency
  4. FDA-ready: Pre-scripted medical protocols are safer than pure generative AI

Demo Best Practices

  • Test 30 minutes before - Load prototype, check browser zoom
  • Have backup - Recorded video if live demo fails
  • Speak slowly - Complex technical content
  • Pause after key points - Let information sink in
  • Face the audience - Not the screen

Handling Questions

  • Technical failures: Use recorded backup
  • Unknown answers: "Let me follow up with you after..."
  • Regulatory concerns: Emphasize FDA Class II pathway with pre-scripted protocols

🔄 Customization Guide

Modify Flight Duration

Edit prototype/dispatcher-console.html:294:

const totalFlightTime = 135; // 2:15 in seconds

Add Voice Conversation

Edit prototype/dispatcher-console.html:276:

const conversationScript = [
    { time: 30, speaker: 'DRONE', text: 'Your message' },
    // Add more...
];

Change UI Colors

Edit mockup .d2 files:

style: {
  fill: "#00AEEF"    // Sky Blue
  stroke: "#39FF14"  // Neon Green
}

Then regenerate: ./generate-mockups.sh

📚 Resources

SkyRescue Links

External Documentation

AI/ML Resources

🤝 Team Collaboration

Software Department Roles

  • Michael Peter - Software (Lead)
  • Shubha - Software

Cross-Team Integration

  • Marketing: Branding, messaging, go-to-market strategy
  • Sales: Customer acquisition, pilot programs
  • Finance: Business model, funding, cost analysis
  • HR: Culture, hiring, remote-first policies
  • Manufacturing: Drone hardware, assembly, QA

Communication Standards

  • Asynchronous First: Slack, Google Docs, Loom
  • AI-Enhanced Meetings: Transcription, translation, summaries
  • Radical Transparency: All decisions documented
  • Outcome-Focused: Measure by deliverables, not hours

✅ Workshop Checklist

Before presenting:

  • All mockups generated (./generate-mockups.sh)
  • Prototype tested in browser
  • API docs viewable in Swagger UI
  • Demo video recorded (backup)
  • Presentation slides prepared
  • Talking points memorized
  • Technical setup tested (screen sharing, zoom level)
  • Handouts printed (API docs, contact info)

After workshop:

  • Follow up with attendees (within 24 hours)
  • Share GitHub repo link
  • Send live demo URL
  • Schedule 1-on-1 for interested investors/partners
  • Measure success metrics (questions, follow-ups, GitHub stars)

🎓 Next Steps

For Development

  1. ✅ Generate and review all mockups
  2. ✅ Test interactive prototype
  3. ✅ Explore API documentation
  4. ⏭️ Integrate with real backend (if building production)
  5. ⏭️ Deploy to staging environment
  6. ⏭️ Connect real drone telemetry

For Workshop

  1. ✅ Practice demo flow
  2. ✅ Memorize key talking points
  3. ⏭️ Prepare backup materials
  4. ⏭️ Test all technical equipment
  5. ⏭️ Rehearse Q&A responses

For Production

  1. ⏭️ Convert prototype to React/Next.js
  2. ⏭️ Implement real WebSocket backend
  3. ⏭️ Add map integration (Mapbox/Google Maps)
  4. ⏭️ Connect to AWS IoT Core
  5. ⏭️ Add authentication and authorization
  6. ⏭️ Implement logging and monitoring

💡 Innovation Highlights

What makes SkyRescue different:

  1. Voice AI on drone - No other AED delivery drone has conversational AI
  2. Edge-first architecture - Runs offline, ultra-low latency
  3. Multi-language auto-detection - Serves diverse communities
  4. Hybrid dialogue system - Safety of scripts + flexibility of AI
  5. Real-time dispatcher visibility - Live transcript and telemetry
  6. One-click launch - AI handles everything after dispatcher decides

Technical achievements in 4 hours:

  • Complete UI mockups with SkyRescue branding
  • Full OpenAPI specification with 15+ endpoints
  • Interactive HTML prototype with live simulation
  • MQTT integration guide with code examples
  • Comprehensive technical architecture documentation
  • Workshop presentation with demo script and Q&A prep

📞 Contact

Software Department Lead: Michael Peter GitHub: https://github.com/michaelryanpeter/CCS-AI-Lab

Company Mission: "First to the scene, first to save"

Built with AI tools in <4 hours 🚀 Demonstrating AI-Powered Velocity: Speed over Perfection

About

No description, website, or topics provided.

Resources

Readme
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

No packages published

Languages

  • HTML 62.7%
  • D2 32.8%
  • Shell 4.5%