Robotics Software Engineer - Drone Autonomy

About the role

Polybee, a Singapore and Melbourne-based startup supported by SGInnovate, builds autonomous drone pollination and yield forecasts for fresh produce. You will join the Product Engineering team as a Robotics Engineer, reporting to the VP Engineering and working on onboard autonomy for the drone fleet. The role is based at the Singapore HQ.

Key responsibilities

• Build and productionize onboard autonomy: localization, mapping and obstacle detection/avoidance so drones navigate safely in cluttered crops and orchards.
• Design and implement visual/inertial localization pipelines such as visual-inertial odometry or SLAM, fusing camera and IMU data.
• Develop real-time obstacle detection and avoidance algorithms within the compute, power and latency budget of the onboard companion computer.
• Integrate autonomy modules with the flight controller and mission-planning stack, working with flight software.
• Own the full pipeline from sensor data capture through algorithm design, on-device optimization, field testing and iteration.
• Partner with computer vision and flight-planning teams to align autonomy behavior with mission requirements.
• Instrument autonomy systems with logging, telemetry and evaluation metrics, and test in simulated and real field conditions.
• Help establish a quality-first code culture: unit and integration tests, code reviews and CI/CD.

Required profile

• 2-4 years building and shipping robotics or autonomy software in production or rigorously field-tested research settings, ideally on aerial or mobile robots.
• Experience building real-time perception pipelines for obstacle detection and avoidance under real hardware constraints.
• Ability to test and validate systems in the field, diagnose failures from telemetry and logs, and iterate quickly.
• Good communication skills and comfort collaborating with computer vision, flight-planning and operations teams.

Required skills

• C++ and Python; computer vision and robotics libraries such as OpenCV, ROS/ROS2 or equivalent.
• Localization and mapping: visual/inertial odometry, SLAM, sensor fusion (Kalman or particle filters).
• NVIDIA Jetson or similar embedded platforms, with algorithm optimization for resource budgets.
• Familiarity with drone flight-control concepts; DJI hardware experience not required.
• Exposure to 3D reconstruction, photogrammetry or structure-from-motion, and experience creating IP, is highly valued.