Job Description

• Contribute to the overall design of various components of our robot perception system (e.g. object detection/tracking, semantic terrain mapping, situation classification, path prediction, structure detection during docking, and anomaly detection). We take the time to ensure everything works robustly and reliably with each sensor/compute configuration.
• Develop/improve the implementation of each step of our sensor data processing pipeline as well documented industry ready software modules and libraries. Currently we use Python (for prototyping) and Rust/C++ (for deployment). Neural networks are deployed using tools like TVM and TensorRT. We develop predominantly for ROS2.
• Design and develop neural network training pipelines based on both real and synthetic data. We have many neural networks deployed on the robot, and these require individual attention to ensure they are working correctly in real word deployments.
• Help build benchmarks to automatically test the performance of the perception system in a range of edge cases.
• Help build tools that allow us to monitor the performance of the perception system when deployed in the field.
• Work closely with other teams (e.g. navigation team and fleet management team) to ensure operational requirements are met.
• Work with other team members with a range of technical roles.
• In addition to the above, you will also be working on system architecture and design, mapping out the technology roadmap and executing them together with the rest of the team.