Robotics Software Engineer

About Omakase Robotics

Aiming to be the world’s leading humanoid company, we are building highly reliable robots for everyday real-world use, engineered with rigorous mass-production discipline. Three vertically integrated components: Omakase D1 (our own hardware), Omakase Zen (the manipulation intelligence foundation), and Omakase OS (the orchestration software that runs robots in the field — this role).

The Role

Omakase OS is everything between a trained model and a robot doing a job in a hospital at 6am: on-robot inference serving, task orchestration and state machines, navigation and manipulation integration, voice/HRI hooks, deployment, monitoring, and recovery. You will build the runtime that makes intelligence operational — the unglamorous, decisive layer where robotics companies live or die.

Responsibilities

Design and build the on-robot runtime: ROS 2 nodes, task orchestration/state machines, safety guards, and recovery behaviors

Serve learned policies on-robot and at the edge (policy servers, streaming inference, latency budgets on Jetson-class hardware)

Integrate subsystems into one working robot: manipulation policies (Zen), navigation/SLAM, voice interaction, fleet telemetry

Own deployment and operations tooling: containerized builds, systemd services, OTA updates, remote debugging of field robots

Build the observability layer: structured logging, telemetry ingestion, incident triage for deployed robots

Work with FDEs to turn field failures into runtime fixes and with the Data Platform team to capture every run as training data

Required Qualifications

5+ years of software engineering , with 2+ years shipping software that ran on physical robots or embedded/edge systems in production

Strong modern C++ (14/17) and Python

Production ROS / ROS 2 experience on Linux (real robots, not only simulation)

Real-time thinking: latency budgets, lock-free queues where needed, graceful degradation, watchdogs

Docker, Git, CI/CD as daily tools; comfort debugging a robot over SSH in the field

Working English (team documentation is in English); Japanese is a plus for field work

Preferred

On-device ML inference (TensorRT, ONNX Runtime, GPU serving) and Jetson/edge optimization

Experience integrating learned policies (VLA/BC) into control stacks

Fleet management, OTA update systems, or remote operations tooling

Voice/multimodal interaction pipelines

Safety frameworks for robots operating around people

Why This Role

OS is where the whole company converges: your code decides whether hardware (D1) plus intelligence (Zen) equals a robot a hospital can rely on.

Field feedback in days: robots you deploy run real shifts at real customer sites in Japan.