Chuanneng Sun

LLM Post Training | Multi-Agent Reinforcement Learning

About Me

My name is Chuanneng Sun. I am an Applied Scientist II at Amazon Stores Foundation AI, where I work on Rufus foundation models, agentic LLM post-training, reinforcement learning, synthetic data generation, and large-scale agent rollouts.

I received my Ph.D. in Electrical and Computer Engineering from Rutgers University, advised by Prof. Dario Pompili. My research focuses on multi-agent reinforcement learning, federated learning, robotics, edge intelligence, and LLM-based decision-making. Before Rutgers, I received my B.S. in Computer Science from Hefei University of Technology.

For more details, please see my Google Scholar or LinkedIn.

News

  • Oct 2026: Our paper Beyond Pairs: Your Language Model is Secretly Optimizing a Preference Graph was accepted to NeurIPS 2026.
  • Oct 2026: Our paper Eluna: An Agentic LLM System for Automating Warehouse Operations with Reasoning and Task Execution was accepted to EMNLP 2026.
  • Oct 2025: Two papers appeared at IEEE MASS 2025: large vision-language model-assisted semantic segmentation and location-/modality-aware heterogeneous data fusion for cooperative perception.
  • Aug 2025: Our multi-personality multi-agent meta-reinforcement learning framework for adaptive transportation systems appeared in IEEE Transactions on Intelligent Transportation Systems.
  • Jul 2025: Our Meta-ETI work on meta-reinforcement learning with explicit task inference for AAV-IoT coverage appeared in IEEE Internet of Things Journal.
  • Jun 2025: Our survey and perspective paper on LLM-based multi-agent decision-making appeared in IEEE Robotics and Automation Letters.
  • May 2025: Our retrieval-augmented hierarchical in-context reinforcement learning work for LLM task planning appeared at IEEE ICRA.
  • Mar 2025: Our heterogeneous federated learning work with generative model-aided knowledge distillation appeared in IEEE Internet of Things Journal.
  • 2024: Received the Best Paper Award at IEEE DCOSS-IoT.
  • 2024: Our contextual biasing work with large language models appeared at ICASSP.

Contact