Research Scientist Intern (TikTok-Neural Graphics and World Models) - 2027 Start (PhD)
Core
Research intern developing foundational neural graphics and world models, focusing on model development, data strategy, evaluation, or interactive prototyping.
Role type
Research Scientist Intern (PhD)
Builds
Research prototypes for interactive graphics, creative tools, or engine-based workflows
Domain
AI for graphics, generative modeling, world models
Deliverable
research
Required skills
machine learning, generative modeling, experimental methodology, PyTorch, JAX, variational autoencoders, latent tokenizers, diffusion models, flow-matching models, diffusion transformers, autoregressive models, multimodal transformers, long-context modeling, pre-training, post-training, 3D modeling, asset generation, 3D representations, animation, motion generation, simulation, physics-aware learning, neural rendering, world models, video prediction, graphics and game-engine workflows
Preferred skills
peer-reviewed publications at leading venues, technical reports, open-source contributions, demos, project pages, 3D or world generation conditioned on text/images/video/actions/structured inputs
Technologies
PyTorch, JAX
Responsibilities
Study relevant research and reproduce baselines, develop and evaluate models/algorithms/data methods, build training/inference/data/evaluation workflows, run controlled experiments and analyze model behavior, create research prototypes, produce reproducible research artifacts and technical reports
Seniority
Intern, PhD candidate
