- Cortex: the engine behind the chat window
7 May 2026
AI has completely changed how I work as an engineer, and we are never going back. This is a look behind the chat window at the platform that made the change possible — two NVIDIA DGX Sparks running vLLM, an OpenAI-compatible proxy with load balancing and full tracing, a vector store with two years of institutional memory, and a fleet of MCP servers wired into the real systems we run on. The model is the smallest piece. Everything else is what makes it useful.
- What an AI engineer actually does in a chat window
6 May 2026
I asked Claude to investigate one customer migration. It pulled data from six systems, wrote a Confluence assessment, opened a Jira ticket, deployed Junos config to two routers, made API changes on the NSX-T fabric, hit a real surprise mid-deploy, recovered cleanly, and posted itself an audit trail. One chat window. A few prompts. This is what AI-driven engineering actually looks like in 2026.
- Claude Provisioned My Homelab Shell VM — in One Conversation
19 Apr 2026
One brief, a handful of skills chained together: how Claude Code stood up a dedicated shell VM for itself, documented it, and handed me the keys.
- Giving Claude Code Long-Term Memory with Qdrant
7 Apr 2026
How I built a semantic memory layer for Claude Code using Qdrant and Ollama, so it remembers past conversations, facts, and decisions across sessions.