2025.11: Published C2RM, a large model RL reward model based on uncertainty modeling, at EMNLP 2025 Main (Oral).
2025.8: Released PiKV, a distributed KV Cache architecture for MoE model optimization.
2025.8: Released VRPRM, a state-of-the-art CoT-PRM model.
2025.8: Released NaviMaster, a unified model capable of operating both digital GUI interfaces and real-world navigation.
2025.7: As Core Lead, responsible for the knowledge enhancement and 'Prudent Mode' modules of SafeWork-R1.
2025.6: Guided an intern to submit a paper accepted by ICCV2025, achieving SOTA results in multimodal retrieval.
2025.5: Published a review article on AI provenance in the Artificial Intelligence Review journal.
2023.10: First author of a paper using dynamic graph network evolution engines for accelerating complex traffic system simulations.
2022.10: Joined Shanghai AI Lab as a young researcher, with excellent performance evaluations for two consecutive years.
2022.11: First-author paper accepted at the first Learning on Graphs Conference.
2022.3: Co-authored a paper published in Nature Machine Intelligence.
2021.9: First-author graph computing paper accepted at SIGMOD 2021 Oral.
2020.2: First paper published in the Knowledge-Based Systems journal.
Research Experience
Joined the Secure and Trustworthy AI Center of Shanghai AI Lab in October 2024, responsible for enhancing the trustworthiness of large models with knowledge. Started working as a young researcher at Shanghai AI Lab in October 2022, mainly focusing on optimizing AI security evaluation systems and multi-agent simulation platforms.
Education
Received a bachelor's degree in Electronic Information Engineering from Sichuan University in 2017; obtained a Ph.D. in Artificial Intelligence from Shanghai Jiao Tong University in 2022.
Background
A young researcher at Shanghai AI Lab, focusing on graph computing and knowledge mining. Currently, his research is centered around developing self-evolving, trustworthy multimodal agents.
Miscellany
Interests include multimodal large language models, self-evolving memory systems, and environment feedback reinforcement learning.