Perception Engineer The RoleAs a Perception Engineer, you'll develop and improve the perception stack that powers our self-driving vehicles. Working hands-on with sonar and image data, you'll help build robust pipelines for detecting and tracking. You'll work closely with the autonomy team to integrate perception capabilities into bespoke behaviors, unlocking novel capabilities.
What You'll Do- Implement and maintain real-time perception pipelines on autonomous vehicle platforms.
- Develop algorithms for obstacle detection, classification, and multi-object tracking.
- Write and maintain ROS 2 nodes and packages for on-vehicle perception software.
- Utilize Python and C++ to develop and implement algorithms for obstacle detection, classification, and multi-object tracking.
- Participate in on-site vehicle testing and debugging of real-world perception failures.
- Collaborate with localization, prediction, and planning teams to ensure clean perception interfaces.
What We're Looking ForRequired- 3-5 years of experience in robotics perception, computer vision, or autonomous systems.
- Strong Python skills for prototyping, scripting, and ML workflows.
- Solid ROS / ROS 2 experience, experience with commercial grade robotic systems.
- Proficient in C++ for real-time, production-grade robotics software.
- Experience with 3D sensor data - LiDAR point clouds, depth estimation, or multi-sensor fusion. Marine sensor experience is a big plus!
- Working knowledge of deep learning frameworks (PyTorch preferred) for detection or segmentation tasks.
- Working knowledge of classical data-driven perception techniques.
- Understanding of 3D geometry, sensor calibration, and coordinate frame transformations.
- Willingness to work on-site and participate in physical vehicle test sessions.
- US Citizen and ability to gain a security clearance.
Preferred- Exposure to Rust for systems programming or robotics tooling.
- Familiarity with AV-specific perception challenges: moving object tracking, occlusion handling, sensor degradation in adverse weather, GPS-denied localization.
- Knowledge of TensorRT or other inference optimization tools for real-time deployment.
- Understanding of AV safety standards or safety-critical software development practices.
Nice to Have- Prior experience at a startup and/or defense company.
- Experience deploying perception models on embedded automotive hardware (Drive AGX, Jetson, etc.)
