大模型应用/agent算法工程师（J101345）
Core
Optimize LLM assistant performance via routing, planning, RAG, and content generation; train models (pretrain, SFT, PPO, DPO, GRPO); and enhance agent reasoning and tool-use capabilities.
Role type
Senior IC large language model application and agent algorithm engineer
Builds
AI assistants with improved response quality, autonomous reasoning, and tool-use skills
Domain
Artificial Intelligence / Large Language Models / Agent Systems
Deliverable
production ML models
Required skills
PyTorch, Transformer architecture, SFT, RLHF, DPO, PPO, GRPO, RAG, Agent frameworks, end-to-end algorithm implementation, data analysis
Preferred skills
Publications in NeurIPS/ICML/ICLR/ACL, Kaggle/ACM awards, open-source contributions, experience in dialogue systems or robot decision-making
Technologies
PyTorch, RAG, Deep Research, RL
Responsibilities
Optimize router and plan strategies for AI assistants; execute multi-stage model training; explore autonomous reasoning and long-term planning; design evaluation metrics and iterate strategies based on data; track and reproduce latest LLM research from top conferences.
