We build.

Assessment and planning are part of the job — you can't change a system you don't understand — but they're the first weeks of a build, not the deliverable. Engaging Longleaf Software means you want an outcome that is real: code that runs, on hardware you have, doing work you can point at.

We work alongside your people

We don't take a specification and disappear for six months. Research software is entangled with the science it serves, and the questions that matter — is this approximation safe to make, does this output still mean what it meant before — can't be answered from a requirements document.

Your researchers stay in the loop. Complex computational problems get solved faster when the people who understand the science and the people who understand the hardware are in the same conversation.

We set the goal first

Every engagement starts by naming the outcome in terms you could verify: this model runs in under an hour, this data is queryable by collaborators at three institutions, this app is in the hands of burn managers before the season opens.

Vague goals produce vague software. If we can't state the target concretely, that's the first problem to solve.

We take on few projects at a time

This work rewards depth. We keep the engagement list short so the person who wrote your kernel is the person who answers the phone about it.

What we won't do

If the honest answer is that a port isn't worth the effort, or that the thing you need is a week of your own developer's time rather than a contract, we'll tell you. We'd rather lose the engagement than build something that doesn't earn its keep.