Forward-deployed engineers

Eighty percent engineering, twenty percent product.

The split is deliberate. Enough product judgement to pick the right problem, and enough engineering to ship the solution without waiting for somebody else to build it.

~80%Backend and full-stack engineering
~20%Product management and technical consulting
One personChooses the problem and ships it

The engineering eighty

The majority of the time is spent writing and running code. That is what separates this role from consulting: the deliverable is a system in production, not a recommendation about one.

  • Backend services and the data pipelines behind them
  • Full-stack work where the interface is part of the problem
  • Model integration, evaluation, and serving
  • Live debugging, performance work, and operational readiness

The product twenty

The remaining time decides whether the engineering was pointed at the right target. It shows up as requirements discovery, scoping, trade-offs explained in your terms, and the honest answer when the answer is "this will not pay for itself".

  • Requirements discovery and scoping
  • Technical consulting, in language your operators use
  • Trade-offs stated plainly, including the ones we dislike

The two failure modes it avoids

Pure consulting produces a strategy that nobody implements. Pure engineering produces a system that is technically correct and answers a question nobody asked. The hybrid split is the cheapest way we know to avoid both.

The rest of the role

The other two parts.

The role is one job with three faces — read the other two, or start from the AI page.

Start with an engineer on site.

Tell us the process that is not working and the systems it touches. We will tell you whether it is worth building, and roughly how long it takes.