微信搜索-视频搜索算法高级工程师-生成式召回
Core
Optimize end-to-end video search algorithms, specifically focusing on generative recall using large language models to enhance user experience and retention.
Role type
Senior IC machine-learning engineer (generative search)
Builds
Next-generation intelligent video search paradigm
Domain
Video search, Generative AI, Large Language Models
Deliverable
production ML models
Required skills
C++, Python, PyTorch, Reinforcement Learning (PPO, DPO, GRPO), Large Language Models (LLM), Encoder-Decoder architectures, Supervised Fine-Tuning (SFT), Semantic ID (SID) clustering and quantization, Multi-modal semantic understanding, User behavior modeling
Preferred skills
Practical experience in generative retrieval or generative recommendation systems
Responsibilities
Optimize full-link video search algorithms including recall, ranking, and mixed ranking; Explore deep integration of large models with massive data to build efficient generative recall; Model user long-term and short-term behavioral interests; Optimize model ability to autoregressively generate target semantic IDs based on search queries and user behavior sequences.
