The engineer who learned that the right decision matters more than the right code.

Early on at Iterpro, the product backlog was basically a Slack thread and a spreadsheet nobody trusted. Every feature request came in with the same label: "urgent, clients are asking for it." Nobody was lying — clients were asking. But nobody had actually verified whether solving that request would move any number that mattered.

So we built a lightweight validation layer: a few questions every request had to answer before it touched the roadmap. Usage data, affected segments, revenue impact, cost to serve without it. Within two months, half the "urgent" requests dropped off on their own. The other half got prioritised faster than ever, because now everyone agreed on why.

That's the work I find most interesting: not writing code, but giving teams the analytical scaffolding to make decisions they can defend.