光子 AI-3D 内容合成数据研究员
Core
Build synthetic 3D content generation pipelines using game engines to produce multi-view synchronized data for AI training.
Role type
Senior IC research engineer (synthetic 3D data generation)
Builds
Procedurally generated 3D scenes, multi-view synchronized datasets (GBuffer, semantic segmentation, optical flow), and 3D asset management systems.
Domain
Computer Graphics / 3D Vision / AI Data Generation
Deliverable
production ML models
Required skills
UE5/Unity development, C++/Blueprint scripting, rendering pipeline expertise, procedural content generation (PCG), Python, PyTorch, USD/Houdini/Omniverse
Preferred skills
Differentiable rendering, sim2real domain randomization, diffusion models, top-tier conference publications in graphics or 3D vision
Technologies
UE5, Unity, Houdini, Blender, USD, Omniverse, PyTorch, Python
Responsibilities
Develop end-to-end synthetic data production pipelines including scene configuration, camera trajectories, and lighting; Implement procedural content generation for batch scene creation; Export engine data including GBuffer, semantic segmentation, and optical flow; Optimize batch rendering efficiency and establish asset reuse mechanisms; Track industry practices and explore sim2real domain randomization applications.
Seniority
Senior, hands-on IC
