多模态大模型算法工程师（J83573）
Core
Researching breakthrough multimodal large model architectures and optimizing training strategies for visual-language-audio-3D fusion.
Role type
Senior IC multimodal large model algorithm engineer
Builds
Multimodal large models deployed in products like Baidu Netdisk, Baidu Wenku, TeraBox, and Chengpian
Domain
Artificial Intelligence / Multimodal Learning
Deliverable
production ML models
Required skills
Transformer, CLIP, Diffusion, MoE, knowledge distillation, contrastive learning, prompt engineering, RLHF, model training optimization, theoretical derivation
Preferred skills
Top-tier conference publications (CVPR/ACL/ICML), core contributions to open-source projects
Technologies
Transformer, CLIP, Diffusion, MoE
Responsibilities
Develop new multimodal fusion paradigms; Solve technical challenges in modality alignment and reinforcement learning; Drive technology productization for billion-user products; Track latest research from top conferences; Deepen product value through technical innovation