智能体算法工程师（J102787）
Core
Design and develop core algorithms for general-purpose AI agents, including task understanding, long-horizon planning, reflection, memory mechanisms, and multi-agent collaboration.
Role type
Senior IC agent algorithm engineer
Builds
General-purpose AI agents with tool use, multi-agent coordination, and agentic capabilities
Domain
Artificial Intelligence / Large Language Models
Deliverable
production ML models
Required skills
Machine learning, deep learning, Transformer architecture, LLM principles, PyTorch, SFT, RLHF/DPO, model evaluation
Preferred skills
Task planning, tool use/function calling, RAG, multi-agent frameworks (LangGraph, AutoGen), agent benchmarking
Technologies
PyTorch, LangGraph, AutoGen
Responsibilities
Design and develop core agent algorithms (planning, reflection, memory); Build tool use and function calling systems; Explore multi-agent collaboration architectures; Construct evaluation systems and data flywheels for continuous optimization; Track and implement frontier technologies like Agent RL and Computer Use.
