微信小微-大模型算法研究员-对话方向
Core
Optimize conversational AI capabilities for WeChat's XiaoWei assistant, focusing on chat, role-playing, Q&A, and multi-turn dialogue scenarios.
Role type
Senior Large Language Model (LLM) Researcher (Dialogue)
Builds
Conversational AI features, generative follow-up recommendation systems, self-evolving experience systems, and face memory/persona graph capabilities for WeChat users.
Domain
Consumer Social Media / Generative AI
Deliverable
production ML models
Required skills
Large model post-training, dialogue optimization, Reward Models, automatic evaluation, Agent self-evolution, RAG, long-term memory, multimodal understanding, data construction, A/B testing, PyTorch, Transformers
Preferred skills
Vibe Coding, AI programming tools, engineering implementation
Responsibilities
Build and optimize data pipelines and evaluation tools for dialogue, harness, and memory systems; conduct online experiments to improve model performance; analyze bad cases and reduce hallucinations in persona recognition.
