AIGC大模型算法专家-国际商业化（北京/上海）
Core
Conduct cutting-edge algorithm research and model optimization in generative AI, focusing on image, video, digital human, and multimodal generation to break technical boundaries in quality, controllability, sampling efficiency, and training stability.
Role type
Senior IC AIGC large model algorithm expert (multimodal generation)
Builds
High-quality generative models for images, videos, and digital humans
Domain
Generative AI / Multimodal / Computer Vision
Deliverable
production ML models
Required skills
Diffusion models, DiT/MMDiT, Flow Matching, Autoregressive models, SFT, DPO, RLHF, PPO, GRPO, reward modeling, large-scale data engineering, automatic annotation
Preferred skills
Publications at top conferences (CVPR, NeurIPS, ICML), representative GitHub repos, deep research in world models or unified generation
Technologies
Diffusion, DiT, MMDiT, Flow Matching, Rectified Flow, PPO, GRPO, SFT, DPO, RLHF, Reward Modeling, ImageReward, VideoScore, SemDeDup, MinHash
Responsibilities
Design and improve generative model architectures (Diffusion, DiT, etc.); Build high-quality training data pipelines including cleaning, synthetic data generation, and automatic annotation; Apply post-training optimization methods (SFT, DPO, RLHF, distillation); Research and implement reinforcement learning algorithms for generative models; Track and integrate state-of-the-art open-source works.
Seniority
Senior, hands-on IC
