June 25, 2026

A recurring piece of advice for new engineering managers is to let go of the keyboard. Your job is the team now, the logic goes, and every hour you spend in the codebase is an hour you’re not spending on the people. There’s truth in it — a manager who hoards the interesting work, or who becomes the bottleneck on every review, is failing at the actual job. But I’ve never fully put the keyboard down, and I don’t intend to.

Judgment decays without contact

The hardest part of technical leadership is making good calls with incomplete information: which migration is worth the risk, where a “quick” integration will quietly metastasize, when “we’ll fix it later” is fine and when it’s a trap. That judgment is built from contact with real systems, and it decays surprisingly fast once you stop having it. Staying close to the code — reading a lot of it, writing some of it — is how I keep the instinct calibrated.

It changes how the team hears you

There’s a difference between a manager who asks “can we make this faster?” and one who has felt where the time actually goes. When I’ve spent an afternoon in the same code my team lives in, my questions get sharper and my “no”s get more credible. People can tell when feedback comes from someone who understands the texture of the work versus someone reading a status summary. That credibility is most of what makes the rest of the job — prioritization, recruiting, the hard conversations — actually land.

The trap to avoid

The failure mode isn’t writing code; it’s writing the wrong code as a manager — grabbing the marquee feature, reviewing so aggressively you become a gate, or shipping something only you understand. The point isn’t to out-engineer the team. It’s to stay fluent enough to lead them well, and to keep the part of the job that made me want to do it in the first place.

This site is a small example of that principle. I could have paid for a template. Instead I rebuilt it from scratch — partly to have something to point at, mostly because building things is still how I think.