Now

Updated August 2026 · This is a now page — a snapshot of current focus, refreshed when the focus changes.

Building

The first of two reference builds: an enterprise-grade agent with a properly typed memory layer over lakehouse data foundations — putting the memory-as-data-modeling thesis through a full production-standard implementation, evaluation harness included. The write-up will land on the work page as an architecture case study.

Focusing

I've sharpened this site — and my work — around a single claim: designing agentic AI systems that survive production. The breadth (architecture, data platforms, product, design, teaching) hasn't gone anywhere; it's now doing its proper job as the foundation under that claim rather than competing with it.

Writing

Developing the agent-memory note toward evergreen, and drafting the long-form version of the argument: agent memory is a data modeling problem, and schema decisions dominate mechanism decisions. New notes are planting on the AI-systems side of the garden — evaluation harnesses, agent failure modes, and data foundations for agents are next in the tray.

Want to talk about any of this? Email me.