感知无图端到端算法工程师（J106645）
Core
Develop end-to-end perception algorithms for mapless/light-map autonomous driving to reduce reliance on high-definition maps and support large-scale operations in complex overseas scenarios.
Role type
Senior IC autonomous driving perception algorithm engineer (end-to-end, mapless)
Builds
End-to-end drivable space and guidance path generation, road structure vectorization, and scene semantic understanding for non-structured environments.
Domain
Autonomous driving, computer vision, generative AI
Deliverable
production ML models
Required skills
C++, Python, Linux development, BEV modeling, Transformer architectures, 3D/sparse convolution, Occupancy networks, agent coding, AI code review and optimization
Preferred skills
Diffusion/Flow Matching for trajectory generation, LLM/VLM/VLA training experience, online mapping and road structure perception implementation
Technologies
BEV, Transformer, 3D convolution, sparse convolution, Occupancy networks, Diffusion models, Flow Matching, LLM, VLM, VLA
Responsibilities
Develop end-to-end algorithms for drivable space and path generation; implement real-time perception and vectorization of road structures; integrate multi-modal large models for scene semantic understanding and spatial localization.
Seniority
Senior, hands-on IC
