Multimodal Controllable Video Generation Researcher

L&Q OASIS PTE. LTD.; Singapore; On-site; SGD 120,000 - 160,000; Full time

Job summary

Tencent's Lightspeed Studios research team seeks a PhD-level AI researcher to design multimodal controllable video generation systems. You will explore conditional control, diffusion-based methods, and low-latency inter-frame strategies for collaborative play across game genres.

Collaboration with engine teams and RL post-training will be essential for production-ready results. Applicants should have a PhD in AI-related fields, strong Python/PyTorch skills, and publications in controllable video

Qualifications

• Ph.D. in AI-related fields with focus on CV/Deep Learning.
• Proficient in conditional generation models and multi-player control technologies; familiar with diffusion models.
• Experience in real-time/interactive video generation or game AI is preferred.

Responsibilities

• Design a multimodal controllable generation framework for various inputs (keyboard, controller, trajectory, voice).
• Research high-fidelity generation under sparse control signals and extend conditional control paradigms in video.
• Collaborate with Engine/Agent teams to define standardized interfaces and drive joint optimization with RL after training.

Skills

Python/PyTorch

Multimodal control

Diffusion models

Reinforcement Learning

Education

PhD in AI/Computer Vision

Job description

Tencent's Lightspeed Studios research team seeks a PhD-level AI researcher to design multimodal controllable video generation systems. You will explore conditional control, diffusion-based methods, and low-latency inter-frame strategies for collaborative play across game genres.

Collaboration with engine teams and RL post-training will be essential for production-ready results. Applicants should have a PhD in AI-related fields, strong Python/PyTorch skills, and publications in controllable video