Shengyao Zhuang

Dr. Shengyao Zhuang is an Adjunct Lecturer at the School of Electrical Engineering and Computer Science, University of Queensland. His research focuses on advancing information retrieval systems, particularly leveraging neural networks, large language models (LLMs), and adversarial machine learning to enhance robustness and efficiency. He specializes in dense retrievers, query processing with typos, and federated search frameworks. His recent work explores vulnerabilities in retrieval systems, cross-modal applications, and optimization of retrieval models through techniques like Matryoshka training and reinforcement learning.

Dr. Zhuang is affiliated with the IELAB research group, evident in collaborative TREC track submissions. His publications span conferences such as SIGIR and ACM venues, addressing challenges in zero-shot search, pseudo relevance feedback, and multimodal document processing.

Notable contributions include developing the Tevatron toolkit and investigating the impact of adversarial attacks (e.g., pixel poisoning) on retrieval systems. He has also explored environmental considerations of IR models, emphasizing sustainable computing practices.