Writing / Architecture and systems
Notes from production.
Architecture, production reliability, and AI-assisted engineering — written from the systems side.
How I use AI agents without outsourcing engineering judgment
A practical AI-assisted architecture workflow using Codex, Claude Code, independent reviewers, evidence, and explicit human decision points.
A practical AI coding workflow with Codex, Claude Code, and reusable skills
How reusable skills, repository guidance, specialized agents, and review gates make AI-assisted engineering repeatable rather than improvisational.
From demo to production: what usually breaks first
A practical map of the first reliability, workflow, and observability gaps I look for.
The backend is where product decisions become real
Why unclear product behavior eventually shows up as queues, retries, state, and edge cases.
What I want to see before I trust an AI workflow
Evals, observability, review points, and the places where human judgment should stay explicit.
Discipline beats talent in production work
Why reliable systems come from boring loops: clear boundaries, observability, review points, and follow-through.
Technical cofounder, founding engineer, or software architect?
A practical guide for founders choosing between a technical cofounder, a founding engineer, and a fixed-scope software architecture engagement.
What a technical cofounder should decide before the first engineering hire
The product, architecture, delivery, and ownership decisions a technical cofounder should make before growing an early engineering team.
Architecture review before scaling a connected product
What to review across devices, mobile applications, cloud services, telemetry, and operations before a connected product scales.
A backup domain is not a backup path
What a domain reputation incident revealed about DNS, TLS, authentication, CORS, service routing, CDN behavior, and production resilience.