Staff AI Software Engineer (Infrastructure, Tooling)

Dotmo is a frontier generative AI company building foundational models. We move fast, and the infrastructure and tooling everything else depends on is built and run by a small number of people. We’re looking for an experienced independent engineer to provide software engineering services in this area.

The engagement is generalist in scope: you’ll design, build, and operate whatever is required within the agreed scope of services — services, pipelines, internal tools, developer and research infrastructure. Range matters here more than any single specialty.

What the engagement is

● System design. You decide how things are put together, where the boundaries go, and what happens when they fail — before anything gets written.

● Reliable systems. Things that keep working unattended, fail loudly rather than quietly, and can be restored when they don’t.

● Orchestrating agents. Most of the code here is written by agents. Your responsibility is to decompose work into pieces they can do well, set the constraints they work inside, verify what comes back, and own every line that ships.

● Knowing where determinism belongs. Consequential decisions stay in code you can test. Model output gets checked, not trusted.

● Whatever needs doing. The scope is broad and may include migrations, test maintenance, cleanup, and other technical work as project needs evolve. You will independently determine the appropriate approach and tools for completing this work.

Who we’re looking for

You’ve built large, mission-critical projects and carried them yourself. Once a problem is defined you take it from there without supervision, and you own the outcome rather than the delivery. Your technology choices are deliberate and you can explain them. You own software and infrastructure end to end, including what happens after launch: you see technical debt forming and deal with it rather than routing around it.

● 6+ years building production software — large, mission-critical systems you’ve designed, shipped, operated, and maintained through several rounds of change.

● Strong computer science fundamentals and coding skills — agents will code for you, but you need to know how to do it yourself.

● Strong Python, SQL and data modeling, and the infrastructure around them: Docker, Kubernetes, CI, secrets, access control.

● Applied LLM engineering: structured output and tool use, writing the prompts, tools, and skill files that make agents reliable at a task, evaluation, guardrails.

● Rigorous testing: pytest, integration tests against real dependencies, and a suite you actually trust.

● Comfort in unfamiliar territory — mobile, desktop, media pipelines, whatever comes up. You don’t need library-level expertise in each. You need to design it, direct agents through it, and recognize when it’s wrong. You’ve already taken a project into a stack you didn’t know and shipped it.

● A record of dealing with technical debt on your own systems before it became a problem.

Who this isn’t for

● You’d rather write the code yourself. Most of the code here is written by agents. You need to be able to write it — that’s why it’s in the requirements above — but if you think models aren’t good enough to be trusted with it, you’ll be working against how we do things every day.

● You’re committed to a particular stack. We’re a research-driven company: the right tool for a problem is often one we haven’t used before, or one nobody has used for this.

● You prefer to work only on problems you choose. The work is driven by what the project needs, and may include whatever technical problems need to be solved.

● You prefer a fixed scope and a steady roadmap. Priorities here follow the research, and they can change. The scope is wide, the pace is demanding and stays that way.

● You need QA, design, and product around you. You define the problem, build it, test it, and judge whether it’s good enough yourself.

Interview

3−4 rounds: system design, coding, data modeling, and working with agents.