Sr. Research Engineer/Scientist (all levels), World Models

Work on TikTok’s Vision-Applied Research team to build next-generation world models. Develop methods and infrastructure to train large-scale generative models using simulated and real-world multimodal datasets, with a focus on long-horizon temporal consistency, realistic physics, and complex object dynamics. Create interactive multi-modal data generation and long-context training pipelines for video generation, enabling real-time user and agent interaction with the model.