仿真世界模型工程师（J97892）
Core
Design and train generative video/multimodal world models for autonomous driving scenarios to predict future image frames and sensor states.
Role type
Senior IC machine-learning engineer (world models & 3D reconstruction)
Builds
Generative video models, 3D scene representations, and simulation environments for autonomous driving
Domain
Autonomous driving, computer vision, 3D reconstruction
Deliverable
production ML models
Required skills
Generative video models, world model architectures, spatiotemporal modeling, 3D scene understanding, neural simulation, PyTorch, large-scale distributed training
Preferred skills
Behavior modeling, traffic scenario generation, closed-loop training systems, algorithm deployment optimization
Technologies
3D Gaussian Splatting, NeRF, PyTorch
Responsibilities
Design and optimize large-scale dynamic scene representation and reconstruction algorithms; combine 3D expressions with generative techniques to build physically consistent world models; implement closed-loop training systems for continuous iteration; improve simulation module efficiency to reduce cloud deployment and operational costs.
Seniority
Senior, hands-on IC