CV

Lead Developer

OpenMOA Project

• Architected and developed a Python library for practical online learning tasks.
• Extended the CapyMOA framework to support data streams with evolving feature spaces.
• Implemented efficient algorithms to handle non-stationary data environment challenges.
• Software available at: https://openmoa.net

Research Lead

LLM-based Synthetic Tabular Data Generation

• Conducted a systematic review of LLM-based techniques for generating high-fidelity tabular data.
• Reproduced representative SOTA algorithms to evaluate performance on diverse datasets.
• Investigating novel methods to bridge the gap between LLM capabilities and structured data utility.

Ph.D.

College of William & Mary

Data Science

• Advisor: Prof. Yi He
• Affiliated with the School of Computing, Data Sciences, and Physics (CDSP).
• Core research focus on utilitarian online learning and data stream evolution.

B.E.

Southwest University

Computer Science and Technology

• Honors: University-level First-class Scholarship, University-level Second-class Scholarship.

Dynamic Mechanism Fidelity for Nonstationary Tabular Data: Evaluation and Mechanism-Guided Generation

AAAI Conference on Artificial Intelligence (AAAI)

Under review. Introduces Dynamic Mechanism Fidelity (DMF) for evaluating nonstationary tabular generation, and DyMeG, a mechanism-guided generator that routes each generation step through shared or shifted low-rank adapters.

OpenMOA: A Python Library for Utilitarian Online Learning

Journal of Machine Learning Research – Machine Learning Open Source Software (JMLR MLOSS)

Under review. A Python library for practical online learning tasks, extending the CapyMOA framework to support evolving feature spaces and non-stationary data streams.

Online Learning from Irregular, Drifting, and Heterogeneous Streaming Data

IEEE/CAA Journal of Automatica Sinica (JAS)

Under review.