Notes, reflections, and career updates.
AI can make engineering teams faster while quietly removing the struggle through which they learn. Sustainable adoption requires enough organisational self-control to protect capability as well as throughput.
Culture is shaped less by the values on the wall than by the behaviour leaders reward, the work they make invisible, and the things they quietly allow to continue.
AI doesn't automatically make you better. It can make you much faster, though, and those are two disparate statements that people assume overlap right up until they realise how different they are.
Inherited systems often look messy, but before rewriting, understand the constraints, scars, and trade-offs that made them work in production.
Delivery improves when teams can surface problems together, finish the improvements they agree to, help each other when work gets stuck, and care enough to fix what slows everyone down
AI agents should be treated like any other actor in a system: give them their own identity, scoped permissions, audit trails, and a deliberate recovery path.
Ambitious career goals are useful when they give direction, but growth comes from the next small, evidence-building step and regular course correction.
If someone cannot leave without damaging the team, I have failed both them and the team.
AI helps me draft, edit, design, and test this site, but the judgement, experience, standards, and responsibility for what gets published remain mine.