Software Builder

Why this job is exciting We're changing how we build. We're moving from teams where each developer owns a slice of an epic to builders who own the whole thing. A builder takes an idea and ships it: designed, implemented, evaluated, deployed. One person, full ownership, from first commit to production. Agents made this possible. At Innovatrics, engineers develop with AI agents as the default way of working: you direct the work, the agents produce much of the code, and you own the result. Claude Code is our standard. The craft is judgment: knowing what to delegate, what to verify, what to build by hand, and how to prove the result is correct. That last part is the heart of this job. We build biometric identity systems used by governments, banks, and security agencies: face recognition, fingerprint matching, identity verification at airports, at borders, in banking apps. This domain is regulated and adversarial. People probe our systems to defeat them. So we ship what we can stand behind, and proving correctness is part of the craft here, with the same weight as writing the code. Where this sits These roles map to the Software Engineer and Senior Software Engineer bands of our career framework. We'll calibrate level together during the process, based on the scope of ownership you've actually carried. The framework runs to Distinguished Principal, and the growth path is real. What you'll be doing Owning features end to end: design, implementation, evaluation, deployment. Done means running in production. Directing AI agents through most of the build, then verifying the output as if you wrote every line yourself. You own agent output the same way you own your own code. Writing evals as part of delivery, so correctness is demonstrated rather than assumed. Improving the shared machinery: when an agent makes the same mistake twice, the context, prompt patterns, or harness gets fixed so it can't recur, for the whole team. Making deliberate calls on when AI is the wrong tool: security-sensitive code, novel biometric algorithms, deep-system work. Working across the stack your product needs, going wherever the problem is rather than waiting at a layer boundary. Collaborating with product and other engineers when you need input, without waiting for permission to move. Requirements We hire builders at 2 levels. The core is the same at both; the scope of ownership differs. A developer who has shipped real things to real users, and can walk us through what broke and what they learned. Comfort across more than one language or stack, with genuine depth in at least one. Builders range wide; the best ones also go deep where their product demands it. Hands-on experience developing with AI agents (Claude Code or comparable), and a clear-eyed view of what they get wrong: plausible-but-broken logic, invented packages, missed edge cases. The habit of verification: you can say how you know a piece of work is correct, whether you wrote it or an agent…