自动标注算法工程师（Tracking方向）_XC
Core
Design and engineer an intelligent automatic annotation system for autonomous driving that fuses detection, segmentation, and multi-object tracking (MOT) algorithms to ensure temporal consistency in video and point cloud data.
Role type
Senior IC multi-object tracking algorithm engineer (autonomous driving)
Builds
High-quality structured data pipelines for perception model training
Domain
Autonomous driving, computer vision, multi-sensor fusion
Deliverable
production ML models
Required skills
Multi-object tracking (MOT/MOTS/3D-MOT), cross-frame ID association, occlusion recovery, trajectory interpolation, ReID, motion modeling, track fusion, detection/segmentation pipeline integration, auto-QA, inference optimization, high-concurrency deployment
Preferred skills
3D tracking, multi-sensor fusion (Camera/LiDAR/Radar), YOLO/DETR/Mask2Former, model quantization, multi-GPU parallel deployment, academic publications (CVPR/ICCV/ECCV/NeurIPS)
Technologies
DeepSORT, ByteTrack, BoT-SORT, OC-SORT, CenterTrack, FairMOT, TrackEval, YOLO, DETR, Mask2Former
Responsibilities
Develop and optimize MOT algorithms for video and point cloud data; Design core algorithms for cross-frame ID association and trajectory interpolation; Integrate Tracking modules with Detection/Segmentation outputs; Build Auto-QA and anomaly detection modules; Optimize inference performance for high-concurrency production deployment; Collaborate with perception and simulation teams to standardize the annotation system
Seniority
Senior, hands-on IC