AI Perception Engineer

SOC classification is a genuine toss-up: the role centers on training and deploying ML/perception models (pointing to 15-2051 Data Scientists), but also involves substantial systems-level C++/Python engineering and edge deployment (pointing to 15-1252 Software Developers). 15-2051 was chosen as primary because model training, evaluation, and data pipeline work are the dominant described activities.

The Requirements section lists computer vision tools as a group ('such as OpenCV, CUDA, TensorRT, ONNX, ROS/ROS2, PyTorch, TensorFlow, or similar'), framed with 'or similar', meaning the capability is required but any one named tool is interchangeable. OpenCV and PyTorch/TensorFlow are emitted as separate required skills with their closest peers in alternatives; CUDA, TensorRT, and ONNX are also captured individually as required given their explicit listing in the same gating section.

C++, C, and Python are listed together as 'a major language such as … or similar' — one is required; the others are captured as alternatives.

ROS/ROS2 appears in the required tools list; both variants are captured as alternatives of each other.

The 3–5 year range in 'Other Requirements' is treated as a minimum of 3 years (the overall years minimum of 3). The role's scope and experience band support a Mid classification despite the 'AI Perception Engineer' title carrying no explicit seniority label.

The posting requires U.S. citizenship or lawful permanent resident status (Green Card), which is a citizenship/work-authorization gate but not a security clearance — the clearance requirement is set to None.

CNNs, LiDAR, CUDA, TensorRT, ONNX, vision-language models, SLAM, visual odometry, synthetic data pipelines, and active learning all appear under 'Ideal Experience' or 'Nice to Have' sections and are marked preferred accordingly.

No compensation figures are provided in the posting.