• Amazon AGI foundation, Seattle, WA.      2024.7-2024.10

Merged large language models by adaptive aligning activation to accelerate online multi-task learning.

• Univeristy of Masssachusetts Amherst, Amherst, MA.      2019.9-2025.6

			Modeling multilingual international news diversity and synchrony at global scope. [Paper]
			Generating Question-answers on large language model to extract Global Disasters. [Paper] [Masslive Coverage]
			Enhancing event discovery via LLM-iterative, cluster-wise data generation on social polls.
			Developing a super large multilingual news frame dataset across media bias levels by human-in-the-loop active learning. [Paper1] [Paper2]

• Univeristy of Californi Los Angeles, California, USA.      2018.7-2018.12

Modeled personalized dynamics of social network and opinion, and predicted evolution. [Paper]

• Zhipu AI (Tsinghua University), Beijing, China.      2018.3-2018.9

			Adaptively collected heterogeneous funding and matched to scholars with decision tree (Aminer online).
			Predicted technology evolution with genetically clustering.   [Code]

• Hunan University, Hunan, China.      2017.3-2017.5

Investigated data visualization field in terms of typical techniques, libraries, softwares, research progress, and open problems.     [ Survey ]

• Central South University, Hunan, China.      2016.3-2017.7

			Devised epidemic routing for vehicular networks with taxi trajectory dataset from MSRA. [ Paper ]
			Devised robust and efficient communication for mobile body area network. [ Paper ]
			Devised adaptive energy equilibrium for e-health network. [ Paper ]

• Briup Technology corporation, Jiangsu, China.     2017.6-2017.7

Developed a smart traffic system on sand table as to route optimization, navigation guidance, driving laws, obstacle avoidance, and flow regulation (server, client, and hardware).

• Central South University, Hunan, China.     2015.7-2015.8

Recognized Chinese speech of surgical terms with Kinect. Built GUI to Visualize objects and actions in voice-activated surgery simulation with DirectUI library.