光子 AI-视频生成内容理解与反馈研究员
Core
Researching AI models to understand, track, and control generated video content, enabling real-time interactive feedback and causal consistency in game scenarios.
Role type
Senior Research Scientist (Generative AI & Video Understanding)
Builds
End-to-end closed-loop systems for controllable video generation and interactive feedback.
Domain
Generative AI, Computer Vision, Reinforcement Learning, Game AI
Deliverable
production ML models
Required skills
Video understanding, Video prediction, Diffusion models, Reinforcement learning, Multi-modal generation, Model architecture design, Training pipeline engineering, Python, PyTorch
Preferred skills
Game AI, Simulation environments, Closed-loop system design, Top-tier conference publications
Technologies
Python, PyTorch, Diffusion models, RL agents
Responsibilities
Analyze semantic, object, and spatial structure of generated content; Implement state tracking and consistency modeling for generated videos; Explore unified generation-understanding model architectures; Research causal consistency control and physical constraints for action-to-video generation; Develop low-latency feedback for real-time interactive generation; Collaborate with Agent/RL teams to build end-to-end control loops; Establish evaluation frameworks for controllable interaction in game scenarios; Define and follow proprietary technical roadmaps.
Seniority
Senior, hands-on IC
