Software Engineer - Data & Sensors

About the role

Here at Humanoid, we believe in a future where robots amplify human potential. That’s why we’ve set out on a mission to build the world’s most capable, commercially-scalable, and safe humanoid robots. We’re bringing that mission to life with HMND‑01 - our rapidly developed humanoid platform being deployed in real industrial environments - and we’re growing the team to take it even further.

ABOUT THE ROLE:

We are looking for a Software Engineer to own the device software and data pipeline for our UMI data collection devices - handheld gripper systems used by human operators to demonstrate manipulation tasks. The data they produce is used to train our robots, so the quality of what you ship determines the quality of robots we deploy.

This is a deeply technical, hands-on role. You will write production-grade device software, build reliable sensor fusion and recording pipelines, and own the validation tooling that keeps our training datasets clean. You will work closely with hardware, AI, and data collection teams.

WHAT YOU’LL DO:

- Develop and maintain device software across the full episode lifecycle: session initialisation, per-sensor stream management, episode recording, and graceful shutdown

- Integrate and synchronise sensor streams - wrist and head cameras, 6-DOF pose trackers, gripper aperture encoders, and optionally force/torque sensors - into temporally-aligned, policy-consumable trajectories

- Define and maintain the data contract between UMI devices and the AI team: stream formats, episode packaging, metadata schemas, and the interface between raw recordings and training pipelines

- Build and maintain data validation tooling that detects dropped frames, timing jitter, pose tracking loss, image quality degradation, and sensor faults before corrupt episodes enter the training dataset

- Own operator-facing device software: status indication, start/stop/cancel episode controls, and clear feedback when a session needs to be retried

- Support field data collection sessions - deploying devices to operators, validating data quality in situ, and iterating quickly when issues surface during collection

- Debug integration failures on device hardware: camera driver issues, bus bandwidth saturation, tracker calibration drift, compute thermal throttling, and storage write failures mid-episode

WE’RE LOOKING FOR:

- Experience in embedded or device-level software in a robotics or sensing context - writing code that runs on the device, not just talking to it

- Proficiency in C++ and/or Python for real-time sensor drivers, inter-process communication, and data recording pipelines

- Solid understanding of multi-modal sensor synchronisation - timestamps, hardware triggers, clock drift - and the practical consequences of getting it wrong for downstream ML

- Familiarity with camera pipelines and the difference between what a sensor reports and what actually lands on disk with correct timing