The instinct in the first 90 days is to write code — after all, that’s what you were hired, or funded, to do. The better use of that time is almost always deciding what not to build yet.
The engineering cofounder’s real first task is drawing the line between what needs to be correct and what only needs to exist. Getting that line wrong in either direction is expensive: over-engineer the wrong part and you burn runway on abstractions nobody uses; under-engineer the wrong part and you spend month four rebuilding the one system the business actually depends on.
Ninety days in, the win isn’t a polished product. It’s a codebase small enough that the whole team can hold it in their heads, and a founder who trusts that the technical decisions being made are the right ones for where the company actually is.