算法工程师（影像/轨迹挖掘方向）-地理位置中台
Core
Develop high-precision algorithms for road sign detection, tracking, classification, and 3D vectorization using vehicle-mounted image/video data, leveraging VLMs and OCR.
Role type
Senior IC computer vision engineer (autonomous driving perception)
Builds
End-to-end automated annotation pipelines and 3D vectorized map representations for road signs
Domain
Autonomous driving / Computer Vision / Geospatial mapping
Deliverable
production ML models
Required skills
CNN, Transformer, ViT architectures, STR/OCR algorithms, Python, C++, PyTorch, model quantization (TensorRT/ONNX)
Preferred skills
BEV perception (e.g., BEVFormer), end-to-end vectorization mapping (e.g., MapTR), VLM/SAM fine-tuning, automated annotation system development
Technologies
PyTorch, TensorRT, ONNX, BEVFormer, MapTR, SAM
Responsibilities
Research and develop algorithms for road sign detection, tracking, and classification; Project 2D road signs to 3D space using BEV and multi-view geometry; Develop algorithms to parse topological relationships between signs and road lanes; Build automated annotation pipelines and change detection algorithms
Seniority
Senior, hands-on IC
