Perception Engineer: Object Modeling

Responsibilities

• Post-process outputs from multiple deep learning networks across diverse sensors (e.g., LiDAR, cameras, radar) to model both dynamic and static objects.
• Track dynamic objects and estimate their kinematic states (e.g., velocity, acceleration, trajectory).
• Design algorithms and metrics to quantitatively evaluate the perception system’s effectiveness (e.g., precision, recall, latency).
• Optimize system performance through iterative improvements based on real-world road test feedback.

Qualifications

• Strong foundation in data structures, algorithms, and kinematics/computer vision principles.
• Proficiency in C++ programming for real-time systems.
• Expertise in at least one of the following domains:
• Point Cloud Perception: Expertise in 3D point cloud processing, including clustering, denoising, registration, object detection, and semantic segmentation.
• Machine Learning: Familiarity with classical algorithms (e.g., Linear Regression, Boosting, Bagging, PCA) and their theoretical underpinnings. Hands-on experience with popular machine learning libraries.
• Performance Optimization: Hands-on experience with CPU/GPU acceleration (SIMD, multithreading, CUDA/OpenCL) and high-performance computing for large-scale data.
• Graphics/Computational Geometry: Experience with computational geometry algorithms (e.g., mesh processing, convex hulls) or 3D graphics optimization techniques.
• Post-Processing: Expertise in multi-object tracking (MOT), sensor fusion, motion modeling, and uncertainty estimation for perception systems.
• SLAM: Familiarity with visual/LiDAR SLAM pipelines (e.g., feature matching, bundle adjustment) and optimization frameworks (e.g., nonlinear least squares, Kalman/particle filters).

Benefits

We provide a competitive compensation package, opportunities for professional growth, participation in a discretionary equity incentive plan, and access to a comprehensive company benefits program, subject to eligibility requirements.