C++ Systems Engineer, Foundations

NEO is a humanoid robot that walks into homes, learns from video, and works alongside people. We need engineers who can make its brain and nervous system faster, more reliable, and more capable.

You'll own the core C++ infrastructure that everything else runs on: sensor pipelines, motor control loops, inter-process communication, and the foundational libraries that ship on every robot. Your code executes thousands of times per second on hardware moving through the real world.

• Design and implement low-latency, high-throughput systems in modern C++ that run reliably on resource-constrained hardware

• Own critical real-time infrastructure: sensor fusion pipelines, motor control interfaces, deterministic scheduling, and the foundational libraries shipping on every robot

• Build the IPC and serialization layers connecting perception, planning, and actuation—lock-free queues, memory pools, shared-memory buffers, and high-rate logging that keeps data flowing without surprises

• Establish performance as source of truth through microbenchmarks, flamegraphs, perf counters, regression tracking, and tooling that catches drift under load

• Debug difficult issues across layers: toolchain/ABI problems, linking issues, kernel interfaces, cross-compilation, on-target performance

• Make the system reliably boring: deterministic behavior where needed, clean failure modes, sharp interfaces other teams can build on

• Ship code that runs on thousands of robots in homes and warehouses, not just prototypes in the lab

• Have 5+ years shipping production C++ in performance-critical systems (robotics, autonomous vehicles, HFT, game engines, OS/runtime, compilers, HPC)

• Think in cycles, cache lines, and budgets—latency, memory, bandwidth, startup time—and design systems that stay within them under load

• Be at home in a Unix/Linux environment, e.g. shells, GNU tools, Unix philosophy

• Have shipped real-time systems where failure has physical consequences

• Debug at the systems level with benchmarking, profiling, tracing, inspecting assembly code, and heavy use of linux kernel interfaces

• Write code other engineers want to build on: clear interfaces, minimal dependencies, obvious correctness