1、Develop LLM-driven agent frameworks for task decomposition, tool calling, and long-term memory management. 2、Explore multi-agent collaboration and adversarial mechanisms to improve group intelligence coordination and game theory abilities in complex environments. 3、Build closed-loop systems integrating reinforcement learning (RL) and world models (World Model), enabling agents to autonomously learn and optimize strategies. 4、Drive technology applications and validation in automated research, intelligent operations, robot control, virtual assistants, etc.

1、PhD or excellent Master’s in computer science, AI, machine learning, or related fields. 2、Familiar with LLM-Agent architectures (e.g., ReAct, AutoGPT, Voyager, OpenDevin) and task planning principles. 3、Proficient in reinforcement learning, sequential decision-making, knowledge memory, and context management algorithms. 4、Experience with PyTorch/TensorFlow, agent framework or multi-agent system development preferred. 5、Strong interest in reasoning, action, and multi-modal interaction, with top conference papers or open-source project experience a plus.