Essays on engineering delivery
Long-form pieces on the systems behind predictable shipping — metrics, bottlenecks, WIP, and the operator practices that move them.
Code entropy: how CI checks keep AI from piling up legacy
Code entropy grows on its own, and AI coding agents speed it up. Seven levels of CI checks that stop legacy from piling up, with tools for Python, PHP, and JS/TS.
AI coding agents: give them the periphery, keep humans on the core
AI coding agents are brilliant at some work and dangerous at other work. A fractional CTO's rule for what to delegate — give them the periphery, keep humans on the core — with the enterprise access problem and where agents actually earn their place.
AI made starting work nearly free. Finishing is still expensive.
Why AI-assisted teams start more work than ever and ship about the same: waiting queues, context limits, and Little's Law applied to agents. A fractional CTO's case for one-thread-one-task and WIP limits in the AI era.
Vibecoded MVP stopped shipping? Manage your AI like a junior
A vibecoded MVP stalls the way junior-built codebases always have: code ships fast and nothing catches mistakes. A fractional CTO's playbook for managing AI coding agents with junior-developer practices — with a real rescue case and numbers.
Average lead time is lying to you — how to forecast delivery with percentiles
Average lead time hides what your delivery system actually does. A fractional CTO's guide to median, spread, and p95 forecasting — how to commit to delivery dates you can hit instead of estimates that keep slipping.
How to diagnose where engineering delivery is actually broken
A fractional CTO's metrics playbook: decomposing T2M into Lead Time and Cycle Time, the role of WIP, Little's Law, and how to introduce WIP limits without team revolt.