In this role, you will define and build the next generation of intelligent 3D content creation systems. You will not only develop core algorithms for generating high-quality 3D assets but also design AI Agents capable of understanding complex user instructions, planning workflows, and autonomously manipulating 3D environments.

Key Responsibilities

• Algorithm Development: Research, design, and implement cutting-edge algorithms for text-to-3D, image-to-3D, and video-to-3D generation. Focus areas include NeRF, 3D Gaussian Splatting, geometry processing, and diffusion models.
• AI Agent Design: Develop "3D Agents" powered by Large Language Models (LLMs) that can understand semantic instructions to retrieve, generate, edit, and arrange 3D assets within a scene autonomously.
• Pipeline Engineering: Bridge the gap between research prototypes and production-ready tools. Optimize generation pipelines for inference speed, mesh topology quality, and texture resolution.
• Innovation & Publication: Keep up-to-date with the latest academic progress. Reproduce SOTA methods and conduct original research. Publishing your findings in relevant conferences or journals is strongly encouraged and supported.
• Collaboration: Work closely with technical artists, graphics engineers, and product managers to integrate generative capabilities into existing 3D workflows (e.g., Unity/Unreal/Blender plugins).

Basic Qualifications

• Education: Master’s or PhD degree in Computer Science, Electrical Engineering, Mathematics, or a related field.
• Research Background: Proven track record of publications in computer vision, computer graphics, or machine learning.Note: While top-tier venues (CVPR, ICCV, ECCV, SIGGRAPH, NeurIPS, ICLR) are a plus, we value high-quality research and distinct contributions in any relevant peer-reviewed venue.
• Technical Stack: Deep understanding of Deep Learning frameworks (PyTorch or TensorFlow) and strong programming skills in Python.
• Fluent Chinese communication skills enabling close collaboration with China-based teams