Active Use Cases - Fako Cluster in Action
Table of Contents
This section documents real-world use cases and practical applications of the Fako cluster services. Each use case includes detailed walkthroughs, screenshots, and implementation details.
📋 Table of Contents
AI/ML Use Cases
- Personal AI Assistant
- Voice-Controlled Home Automation
- Document Analysis Pipeline
- Multi-Model AI Comparison
Development & Automation
Personal Productivity
Infrastructure Management
🤖 Personal AI Assistant
Overview
A comprehensive AI assistant setup using Open WebUI, Ollama, and GPUStack for various AI models.
Components Used
- Open WebUI: User interface
- Ollama: Local model hosting
- GPUStack: GPU acceleration
- PostgreSQL: Conversation storage
Implementation
[Detailed implementation will be added here]
Screenshots
🎤 Voice-Controlled Home Automation
Overview
This implementation demonstrates a complete voice assistant pipeline using Wyoming Satellite with Home Assistant, achieving a successful voice interaction where a user queries smart home status and receives an AI-generated response in 17 seconds.
Components Used
- OpenWakeWord: Wake word detection (“Hey Jarvis”)
- Whisper: Speech-to-text transcription
- Ollama (Llama 8B): Natural language understanding and response generation
- Piper TTS: Text-to-speech synthesis
- Home Assistant: Smart home orchestration and pipeline management
- Wyoming Satellite: Voice satellite software on Raspberry Pi
Hardware Setup
- Raspberry Pi 5 (2.4GHz, 4GB RAM) - Voice satellite device
- ME6S USB Microphone - Audio input device
- Bluetooth Speaker (88:92:CC:5D:CB:79) - Audio output via BlueALSA
- Mini PC - Home Assistant server
- Kubernetes Cluster - Container orchestration for AI services
Figure 1: Overall voice assistant architecture showing component interconnections
Implementation
Example Interaction
User Query: “Any lights on in the master bathroom?”
System Response: “Based on the live context, there are no lights currently on in the master bathroom. The Master Bathroom Lights are off.”
Total Time: 17 seconds
Pipeline Stages
Figure 2: Complete voice pipeline flow from wake word to response
1. Wake Word Detection (0-1 second)
- Microphone continuously captures audio at 16kHz
- Audio streams to Wyoming Satellite service
- Satellite forwards to OpenWakeWord pod (K8s NodePort 30400)
- Detection confirmed for “hey_jarvis” with confidence threshold
2. Speech Recognition (1-4 seconds)
- User speech captured and streamed to Home Assistant
- Audio forwarded to Whisper STT pod (NodePort 30300)
- Real-time transcription returns: “any lights on in the master bathroom”
3. Intent Processing & AI Response (4-7 seconds)
- Home Assistant identifies intent: query light status
- State query retrieves:
light.master_bathroom_lights(state: off) - Context sent to Ollama pod running Llama 8B
- LLM generates natural response with current state information
4. Text-to-Speech Synthesis (7-9 seconds)
- Response text sent to Piper TTS pod (NodePort 30200)
- Speech synthesized using ’en_US-ryan-high’ voice model
- Audio generated at 22050Hz sample rate
5. Audio Playback (9-17 seconds)
- Audio streamed back through Wyoming Satellite
- BlueALSA manages Bluetooth A2DP protocol
- Real-time playback through Bluetooth speaker
Performance Metrics
| Stage | Duration | Percentage |
|---|---|---|
| Wake Word Detection | 1s | 6% |
| Speech Recognition | 4s | 24% |
| AI Processing | 3s | 18% |
| TTS Generation | 2s | 12% |
| Audio Playback | 8s | 47% |
| Total Pipeline | 17s | 100% |
Key Optimizations
- CPU Governor: Performance mode (2.4GHz constant)
- WiFi Power Save: Disabled for consistent latency
- Network Timeouts: Increased to 30s (handles 700ms spikes)
- Audio Buffers: 4096 frames for smooth playback
Network Configuration
_PONG_TIMEOUT: Final = 30 # Increased from 5
_PING_SEND_DELAY: Final = 5 # Increased from 2
_WAKE_INFO_TIMEOUT: Final = 10 # Increased from 2
Success Indicators
✅ Reliable “Hey Jarvis” detection
✅ Accurate speech-to-text conversion
✅ Correct intent extraction
✅ Natural, contextual AI responses
✅ Clear TTS playback via Bluetooth
✅ No disconnections despite 37% packet loss
✅ 17-second end-to-end response time
Data Flow Architecture
Figure 3: Linear representation of the data flow through the system
1. Audio Input: User → Microphone → Pi
2. Wake Detection: Pi → WiFi → K8s Wake Pod
3. Transcription: Pi → HA → K8s Whisper Pod
4. Processing: HA → K8s Ollama Pod (Llama 8B)
5. Synthesis: HA → K8s Piper Pod
6. Audio Output: Pi → Bluetooth → Speaker → User
Figure 4: Simplified sequence diagram showing the interaction between components
Key Achievement
Despite operating over WiFi with significant latency variations (38-700ms) and packet loss (37%), the system maintains stability through optimized timeouts and performance tuning, delivering a reliable voice assistant experience capable of understanding context, querying smart home devices, and generating natural language responses.
Additional Architecture Views
Figure 5: Detailed system architecture showing all components and their relationships
Figure 6: Component-level view of the voice assistant system
📄 Document Analysis Pipeline
Overview
Automated document processing using AI for summarization, extraction, and insights.
Components Used
- n8n: Workflow orchestration
- Ollama: Document analysis
- PostgreSQL: Results storage
- Blog: Publishing insights
Implementation
[Detailed implementation will be added here]
Screenshots
🔍 Multi-Model AI Comparison
Overview
Comparing responses from different AI models for the same prompts.
Components Used
- KAgent: AI agent framework
- Multiple Models: Via Ollama/GPUStack
- Open WebUI: Comparison interface
Implementation
[Detailed implementation will be added here]
Screenshots
🚀 GitOps Workflow Management
Overview
Complete GitOps setup with automated deployments and security scanning.
Components Used
- Gitleaks: Secret scanning
- Kube-bench: Security compliance
- n8n: Automation workflows
- Blog: Documentation
Implementation
[Detailed implementation will be added here]
Screenshots
🔒 Automated Security Scanning
Overview
Continuous security monitoring and compliance checking.
Components Used
- Kubescape: Security scanning
- Kube-bench: CIS benchmarks
- Gitleaks: Secret detection
- n8n: Alert automation
Implementation
[Detailed implementation will be added here]
Screenshots
📊 Health Data Analytics
Overview
Personal health tracking and analytics using Oura Ring data.
Components Used
- Oura Collector: Data ingestion
- Oura Dashboard: Visualization
- PostgreSQL: Data storage
- n8n: Health insights automation
Implementation
[Detailed implementation will be added here]
Screenshots
📚 Knowledge Management System
Overview
Personal knowledge base with AI-powered search and insights.
Components Used
- Linkding: Bookmark management
- Blog: Documentation
- Ollama: AI search
- PostgreSQL: Data storage
Implementation
[Detailed implementation will be added here]
Screenshots
🎬 Media Server Setup
Overview
Complete media server with audiobook support and AI recommendations.
Components Used
- Audiobookshelf: Media server
- n8n: Download automation
- Ollama: Content recommendations
Implementation
[Detailed implementation will be added here]
Screenshots
🖥️ Multi-Node GPU Scheduling
Overview
Efficient GPU resource allocation across multiple nodes.
Components Used
- Node Labeling: GPU node identification
- GPUStack: GPU management
- Ollama: Model deployment
Implementation
[Detailed implementation will be added here]
Screenshots
🌐 Secure Remote Access
Overview
Zero-trust access to cluster services via Cloudflare Tunnels.
Components Used
- GPUStack Proxy: Cloudflare tunnel
- Keycloak: Authentication
- OAuth2 Proxy: Service protection
Implementation
[Detailed implementation will be added here]
Screenshots
💾 Automated Backup Strategy
Overview
Comprehensive backup solution for all cluster data.
Components Used
- PostgreSQL Cluster: Database backups
- n8n: Backup automation
- AWS S3: Offsite storage
Implementation
[Detailed implementation will be added here]
Screenshots
📝 Contributing Use Cases
To add your own use case:
- Create a new section in this README
- Add screenshots to the
images/directory - Include:
- Overview of the use case
- Components used
- Step-by-step implementation
- Configuration examples
- Screenshots
- Lessons learned
🖼️ Image Guidelines
When adding screenshots:
- Use PNG format for clarity
- Recommended size: 1200x800px
- Include captions
- Blur any sensitive information
- Use consistent naming:
use-case-description.png
This documentation is a living document. Please contribute your own use cases and experiences!