Why I Architect And Build, Not Just Advise

A data architecture meets reality in week two, and someone has to adjust it on the spot.

That’s why I stay hands-on in most engagements, YAML and all.

The source system has a field nobody mentioned. The team knows Python and not Scala. The budget line for the streaming tier didn’t make it through finance.

Somebody has to adjust the design on the spot. If the architect has already left, the team adjusts it alone, under deadline, without the reasoning behind the original choice. Six months later the diagram describes a system that doesn’t exist.

When I build alongside the team, the compromises get made by someone who knows which parts of the design are load-bearing and which were a preference. And the team picks up the why while we build. That’s the part I care about most, because the goal is that they run it without me.

It’s slower to sell. A deck is easier to scope than “I’ll be in your repo two days a week.”

Advisory still has its place, once the platform is stable and the team is senior. Before that, I’d rather have my hands in it.

What was the first thing that broke your last data design?