Orbital Industries is an AI Industrial company, with frontier AI embedded at every step in the production of critical physical products - from creating advanced materials to engineering and manufacturing.

Every Orbital Industries product is developed using CurieOS, our AI operating system, uniting AI-automated hardware engineering with AI-designed material science to achieve breakthrough real-world performance. We also offer CurieOS to our customers and partners, extending the same platform and workflows that power Orbital Industries to their own teams and products.

We have an ambitious mission and need excellent people in all our teams - AI research, operations, advanced materials, mechanical engineering, chemical engineering and manufacturing.

Working at Orbital Industries means working in vertically integrated teams across the full stack, from molecules to manufacturing. We're looking for people who have a love of physical technology, curiosity in AI and a desire to learn.

The role

Physical engineering is one of the hardest domains for AI. Designs have to meet standards, calculations must be deterministic and checkable, and a wrong number can end up in a power system or a data centre. CurieOS is built to let agents do real engineering work and to show exactly how every output was produced.

The platform team builds the foundation that every workflow, agent and product on CurieOS depends on, and runs it in production for people who need to trust it. Your work sets the floor for how reliable our AI is, and the ceiling for what it can do.

This is a small team, and you'll shape what we build as well as how. You'll see the results of those decisions in production within days. If you want a narrow, well-defined slice of a large system, this isn't the role. If you want to own a platform while it scales, it is.

How we work

We build with agents, not just for them. Everyone on the team works with the latest frontier models and coding agents every day, and token budgets won't hold you back. Part of what we build is CurieOS's ability to work on itself: triaging defects, proposing fixes and opening pull requests for engineers to review. So engineers here spend more of their time on design, review and judgement, and less on typing. If you already work this way, you'll feel at home. If you want to see how far it can go, this is the place to find out.

You'll also work alongside our AI researchers, turning cutting-edge research and evals into things that run in the platform and in production.

What you'll work on first

In your first six months you'll work across the parts of CurieOS that customers have to be able to trust. Like everyone on the team you'll touch most of them, and go deepest where your interests take you:

Tenancy, isolation and identity. How CurieOS keeps each customer separate, from shared multi-tenant accounts to dedicated single-tenant deployments. That covers data isolation, SSO and access control. You'll also be an engineering voice in customer security reviews.

Release and production. How changes get to production safely: release gates, staged rollouts and rollback, automated smoke tests, on-call and incident response. We ship often, and you'll build what lets us keep doing that without breaking what customers rely on.

Agent quality in production. Measuring how well agents perform on real customer workflows, working with our researchers to bring the latest evals into the platform, and using those measurements to decide what is good enough to release. A model or prompt change should have to pass the same gate as a code change.

Observability and cost. Knowing the platform is healthy and what it costs. That means SLOs, tracing and alerting, plus cost telemetry and usage metering for every workflow and every customer, so we can see where model and compute spend goes and make deliberate choices about it.

Infrastructure and deployment. One standard for how CurieOS is deployed, configured and upgraded in each environment, across our own cloud accounts, GPU capacity and customer-dedicated deployments.