How to evaluate a product engineer through their work ↗
A practical hiring conversation around inspectable artifacts: product decisions, implementation boundaries, operational ownership, and tradeoffs.
A practical guide / Engineering leadership
Connect daily engineering habits to the decisions around them: what to build, how to coordinate, and where a technical leader can remove friction.
Follow the reading path ↓Start here
Read in order, or choose the problem closest to your own.
A documented account of coordinating engineering work with an AI-assisted brief, PR triage, and recurring checks.
How I Run 3 Engineering Teams with 1 AI Brain ↗Capture the context and tradeoffs around an architectural choice, so the team can understand it after the meeting ends.
I Taught My AI Agent How to Write Architecture Decision Records ↗Examine the cost of coordination and decide which conversations help the work move forward.
The hidden costs of meetings ↗The working principles
A technical leader helps a team choose what to do next and understand why. Name the constraint before choosing the tool: an unclear product goal, a system boundary that no longer fits, or a delivery process that hides risk until late in the work.
The architecture decision record guide offers a way to keep that context with the decision. The useful output is not a document for its own sake; it is an explanation the next person can inspect and revisit.
A list of activity is different from an account of what needs attention. Look for work that is blocked, changes that need a decision, and operational problems that keep recurring. Make the ownership and next step clear enough that someone can act.
How I Run Three Engineering Teams with One AI Brain describes a specific operating workflow, including its limitations. I Built a Brain with Claude Code explores the coordination layer in more detail.
A recurring meeting should earn its place by helping people make a decision or move work forward. Inspect the cost of interruptions alongside the value of the conversation. Written context can make the discussion more useful, but it does not replace difficult conversations or clear ownership.
The Hidden Costs of Meetings looks at that tradeoff. Pair it with The 200-Line Rule for a smaller, reviewable unit of engineering progress.
From the idea to the work
Work through architecture, delivery habits, and the engineering decisions that are slowing your team down.
Go further
Further experiments, implementation details, and lessons from the work.
A practical hiring conversation around inspectable artifacts: product decisions, implementation boundaries, operational ownership, and tradeoffs.
A proposed first-month approach to fractional engineering leadership: understand the system, review decisions, agree ownership, and establish a useful cadence.
When I first let AI agents write code autonomously, the PRs were enormous. One constraint changed everything.
One prompt. No additional input. I let Claude make all the decisions — project structure, skills, scheduling, everything planned around my desired outcome. Here's what happened.
This isn't a post about replacing developers. I'm a developer. I like being employed. This is about what happens when you stop treating AI coding tools as fancy autocomplete and start treating them as actual teammates.
What if checkout was not an engineering-owned system at all? What if it was a marketing-owned conversion tool, built on solid engineering foundations? This is the story of how we rebuilt checkout from the ground up. Not as a form. As a platform.
Tired of proposals disappearing into email purgatory? I built a Next.js + Sanity system that transforms static PDFs into interactive web experiences with one-click acceptance, instant notifications, and optional Stripe payments. Here's how it works and why it converts better.
If you're using Sanity Studio for content management, you know how crucial great imagery is for your projects. This plugin brings the power of Google's Gemini 2.5 Flash Image model directly into your Sanity Studio interface.
Discover Sanity Agents - 7 powerful Claude Code plugins that revolutionize Sanity CMS development with AI-powered schema building, GROQ optimization, and zero-downtime migrations.
You start your day excited. Claude Code is going to help you crush this feature. Three hours later, you're staring at TypeScript errors you didn't catch, inconsistent code patterns, and Claude confidently telling you those 47 errors are "unrelated to our changes."
Why we left Shopify behind and started building a custom ecommerce experience with Sanity CMS. This is Part 1 of a technical series covering the full build.
How we built a complete proposal management system with digital signatures, payment integration, email notifications, and a drag-and-drop Kanban board—all powered by Sanity CMS.