Posts tagged "observability"

• Building the AI on the Edge Demo System

  2026-08-31· ai-on-the-edge·Estimated reading time: 8 min read

  Engineering wrap-up for AI on the Edge: repo layout, make demo-laptop through make teardown targets, seed, reset, and smoke tests, shared health endpoints, and a repeatable 45-minute conference flow.

• Specialized Hardware Without an App Rewrite

  2026-08-24· ai-on-the-edge·Estimated reading time: 6 min read

  AI on the Edge Demo 6: registering a Tenstorrent or mock accelerator as an OpenAI-compatible gateway backend with capability discovery, health checks, fallback, and accelerator-preferred routing - no application rewrite.

• When the Edge Has to Stand Alone

  2026-08-17· ai-on-the-edge·Estimated reading time: 7 min read

  Demo 5 of AI on the Edge: disconnected operation, backend failures, malformed responses, cloud-blocked mode, local-only denial, smoke tests, and reset controls.

• Edge AI You Can Actually Operate

  2026-08-10· ai-on-the-edge·Estimated reading time: 8 min read

  Demo 4 of AI on the Edge: operating local AI with Azure Arc, GitOps, Azure Monitor, Managed Prometheus, Grafana, Key Vault, policy events, KQL, and dashboard evidence.

• Private RAG That Cannot Leave the Edge

  2026-07-27· ai-on-the-edge·Estimated reading time: 7 min read

  Demo 2 for AI on the Edge: document ingestion, local embeddings, vector search, citations, data classification, LocalOnly routing, and fail-closed cloud fallback.

• One App, Many Places to Run AI

  2026-07-20· ai-on-the-edge·Estimated reading time: 8 min read

  Demo 1 for AI on the Edge: my .NET gateway, OpenAI-compatible facade, backend registry, route explanations, target backends, and LocalOnly denial behavior.