感知语义理解算法工程师（J106597）
Core
Develop end-to-end vision-language large models, traffic sign/light recognition, 2D/3D object detection, and scene semantic understanding for autonomous driving perception systems.
Role type
Senior IC perception algorithm engineer (vision-language models)
Builds
End-to-end vision-language large models, perception modules for autonomous driving
Domain
Autonomous driving, Computer Vision, Large Language Models
Deliverable
production ML models
Required skills
Agent coding, C++, Python, Linux development, Deep learning (3D/Sparsity/Transformers), LLM/VLM/VLA principles, Mathematical analysis
Preferred skills
Publications in CVPR/ICCV/ICRA/IROS/PAMI, Training experience for LLMs/VLMs/VLAs
Technologies
C++, Python, Linux, Transformers, 3D Convolution, Sparse Convolution, LLM, VLM, VLA
Responsibilities
Research and develop vision-language end-to-end large models; Conduct algorithm research and implementation; Define data requirements and build automated data pipelines; Refactor existing perception modules to create agentic data loops.
