多模态算法实习生（J99677）
Core
Research and develop multimodal algorithms for e-commerce digital human generation, focusing on live streaming and short video creation.
Role type
Research intern (multimodal algorithms)
Builds
E-commerce digital human solutions for live streaming and short videos
Domain
E-commerce, Computer Vision, Generative AI
Deliverable
production ML models
Required skills
Diffusion Models, DMD, GAN, VAE, DPO, LoRA, model distillation, 3D face modeling, talking-head generation, lip-sync, DiT, mmDiT, ControlNet, IP-adapter, CV algorithms (classification, segmentation, super-resolution)
Preferred skills
Top-tier conference publications, Wan, CogVideoX
Responsibilities
Explore self-developed model architectures for digital human generation, validate and implement innovative algorithm solutions, publish research papers, optimize full-link digital human solutions for real-time interaction and generation consistency
