微信搜索-高级研究员-大模型应用搜索方向
Core
Leading the upgrade of WeChat video search ranking systems to a generative AI paradigm, shifting from discriminative modeling to generative understanding of user intent.
Role type
Senior IC Machine Learning Researcher (Generative Search)
Builds
End-to-end satisfaction and relevance modeling for video search, post-training infrastructure for large models, and automated data flywheels for iterative model improvement.
Domain
Consumer Internet / Video Search / Large Language Models
Deliverable
production ML models
Required skills
Generative AI ranking, Reinforcement Learning (RLHF/GRPO/DPO), Distributed training optimization, Search ranking systems, Large-scale inference optimization, Data flywheel design
Preferred skills
Agentic Search, Reasoning capabilities, Generative retrieval
Technologies
Megatron-LM, DeepSpeed, veRL, vLLM, SGLang
Responsibilities
Design and implement generative ranking architectures for video search; Develop end-to-end joint modeling for user satisfaction and relevance; Build post-training systems using SFT and RL; Establish data flywheels for automated annotation and evaluation; Scale frontier technologies like Agentic Search to hundreds of millions of users.
