Writing

Things I know, things I'm learning, and things work has taught me, mostly about getting agent systems into production. Longer essays go on Substack; notes live here.

Five areas: Deployments, Harness & evals, Security, Models, Learning. Titles without links are drafts in progress.

Essays

• Forward Deployed Engineer: What the Title Doesn't Tell You19 Oct 2026 · Substack

Deployments

Getting agent systems into a client's environment: on-prem and hybrid rollouts, legacy integration, what changes once it's live.

• Agentic deploymentsUpcoming

Harness & evals

Orchestration loops, tool interfaces, context assembly, guardrails and fallbacks, and the evaluation and tracing that gate releases.

• Starting here27 Oct 2026

Security

What an agent with tools can do wrong, and how to design so it can't.

• Security in agentic applicationsUpcoming

Models

What the models can and can't do, fine-tuning, and choosing between them under enterprise constraints.

• Nothing here yet

Learning

How to keep learning while shipping, and what I'd tell myself a year ago.

• The learning and shipping tradeoff in AI development todayUpcoming