SMAROBIX seeks engineer to build runtime observability for ROS-based robots

SMAROBIX, a TU Dresden startup founded by a team of five engineers from diverse backgrounds, is building observable, understandable, and actionable tooling for ROS-based robotic systems. They are now hiring a Runtime Systems Engineer to join their core team in Dresden (on-site preferred, hybrid possible within the EU). The role centers on designing and implementing a low-overhead runtime tracing system that captures what happens inside dynamic, multi-process ROS 2 systems under real conditions. Unlike existing solutions, this instrumentation must run directly on production robots with minimal overhead and zero correctness compromise. The engineer will own the critical decisions about what to capture, at what granularity, and how to represent that data so the rest of the product can reason reliably. Prior work with kernel modules, deterministic networking, real-time OS, or embedded hardware acceleration is a plus, but the founders emphasize genuine depth in systems thinking over tool familiarity.

The challenge is novel because ROS 2 middleware is inherently opaque at runtime: nodes appear and disappear, communication crosses process boundaries, and QoS settings affect timing and reliability in subtle ways. The engineer must decide how to instrument without adding latency, how to trace across distributed systems, and how to guarantee correctness under load. The founders describe this as a hands-on, design-and-build role where there is no template to follow. The successful candidate will work closely with the co-founder responsible for runtime architecture, with real ownership and room to challenge assumptions. As the team grows, this role naturally evolves into technical leadership through clarity and influence. SMAROBIX sees this position as foundational for their product's long-term reliability—what the engineer captures becomes the ground truth that every other layer depends on. Applications should include a CV and a cover letter describing a hard technical problem where the applicant recognized their first approach was wrong.