Interactive Vision Model Research Intern 108277
Core
Design and iterate next-generation video generation foundation architecture, focusing on long-context attention, KV Cache compression, and high-compression video tokenizers.
Role type
Research intern (video generation foundation models)
Builds
Next-generation video generation models for games
Domain
AI research, video generation, game technology
Deliverable
production ML models
Required skills
diffusion models, autoregressive generation, PyTorch, large-scale distributed training, video VAE design, tokenizers, curriculum learning, scaling laws
Preferred skills
end-to-end video foundation model projects, core industry product R&D
Technologies
PyTorch, diffusion models, autoregressive models, video tokenizers
Responsibilities
Design and iterate video generation foundation architecture; explore long video generation core technologies; research high-compression video tokenizers; lead video pre-training; explore scaling laws; track state-of-the-art and publish papers
Seniority
Intern, research-focused
