Multimodal Controllable Video Gen Researcher

LightSpeed Studios; Singapore; On-site; SGD 120,000 - 180,000; Full time

Job summary

LightSpeed Studios in Singapore seeks a PhD-level researcher to advance controllable video generation for interactive gaming. You will design multimodal frameworks and pursue high-fidelity results under sparse control signals, collaborating with engine teams to define interfaces for controllable signals.

The role emphasizes research in camera, trajectory, and posture control, real-time performance, and integration with reinforcement learning post-training. Prior game AI experience is a plus.

Qualifications

• PhD in a relevant AI field with strong publication record.
• Proficient in conditional generation models and multi-player/multi-channel control technologies.
• Familiar with diffusion model principles.

Responsibilities

• Design a multimodal controllable generation framework for inputs like keyboard, controller, trajectory and voice.
• Research high-fidelity generation with sparse control signals; extend conditional control paradigms in video.
• Decouple and integrate trajectory, camera, motion, and posture controls.
• Develop dedicated control modules for different game genres.
• Build video generation capabilities for multi-player collaboration with low latency.
• Collaborate with Engine/Agent teams to define standardized interfaces for controllable signals.
• Drive joint optimization of controllable generation and RL post-training.

Skills

PhD in AI/Computer Vision

Conditional generation & diffusion

Multichannel control & gameplay

Camera/trajectory/posture control

Engine I/O & frame sync

Python/PyTorch publications

Real-time video generation

Game AI / game development

Education

PhD in AI/Computer Vision/Deep Learning

Tools

Python/PyTorch

Job description

LightSpeed Studios in Singapore seeks a PhD-level researcher to advance controllable video generation for interactive gaming. You will design multimodal frameworks and pursue high-fidelity results under sparse control signals, collaborating with engine teams to define interfaces for controllable signals.

The role emphasizes research in camera, trajectory, and posture control, real-time performance, and integration with reinforcement learning post-training. Prior game AI experience is a plus.