Mesh: Risk Assessment and Mitigation

1. Local Hardware Constraints (RAM/CPU)

• Problem: Running heavy LLMs locally can strain systems.
• Mitigation:
  • Use quantized models (Gemma 4 E4B fits in 6GB).
  • Use Docker profiles to start Ollama only when needed.
  • Implement memory limits in Docker.

2. Scraping Failures

• Problem: Targets may block requests or change structures.
• Mitigation:
  • Circuit breakers (gobreaker) to prevent runaway failures.
  • User-Agent rotation and respect for robots.txt.
  • Exponential backoff for job retries.

3. Data Privacy and Sovereignty

• Problem: Knowledge graph data is highly sensitive.
• Mitigation:
  • Zero data leaves the machine. All inference is local via Ollama.
  • No cloud telemetry or third-party tracking.
  • All services bind to 127.0.0.1 only.

4. Concurrency and Race Conditions

• Problem: Multiple workers accessing the same resources.
• Mitigation:
  • PostgreSQL Atomic UPSERTs (ON CONFLICT).
  • FOR UPDATE SKIP LOCKED for job queue processing.
  • Optimistic Concurrency Control (version column) on nodes.

5. Scope Creep

• Problem: Side projects often stall due to over-ambition.
• Mitigation:
  • Modular phases where each delivers standalone value.
  • MVP-first focus (Phases 1-4).
  • No hard deadlines; sustainable 6-8 hours/week pace.