Designing a multi-server Model Context Protocol (MCP) control plane to connect autonomous AI agents to homelab Docker hosts, UniFi, NAS, and Home Assistant.
Implementing a self-hosted, privacy-preserving AI web search and Retrieval-Augmented Generation (RAG) pipeline using containerized SearXNG and Open WebUI.
Building a Python middleware proxy on Google Cloud Run for OIDC token injection, payload sanitization, and tool_choice schema translation between Open WebUI, LiteLLM, and vLLM.
Architecting a resilient, stateful autonomous AI agent on serverless Google Cloud Run with GCS state persistence, Telegram bot control, and Vertex AI failover.
Deploying open-weight LLMs in pure BF16 precision on serverless NVIDIA RTX PRO 6000 Blackwell (96GB VRAM) GPUs with vLLM on Google Cloud Run.
How to build a Level 5 Agent-Native website in Hugo: serving clean Markdown endpoints (/index.md), llms.txt, RFC 8288 Link headers, and Model Context Protocol (MCP) discovery.
How to migrate enterprise AI workflows and team chat histories to Google Workspace with Gemini Enterprise while maintaining governance and data security.
Reflecting on 12 years in the VMware vExpert program, Broadcom licensing shifts, and why I pivoted my career toward Google Cloud, Kubernetes, and Enterprise AI architecture.