I’m Xiao Wang, an MPhil researcher in Computer Science at the University of Sydney, focused on medical AI for histopathology vision-language understanding. My first-author work has been published at ECAI, where I investigated how vision-language foundation models can better perform zero-shot tumour malignancy recognition through retrieval-augmented and retrieval-denoising language modeling approaches. I’m also building next steps in my research, including instruction-guided retrieval refinement and knowledge-preserving domain adaptation so models can learn new domains without losing their zero-shot and retrieval capabilities. Beyond research, I’ve contributed as an AI engineer and algorithm engineer—working with real-world sensing and computer vision pipelines—while continuously strengthening my engineering toolkit with PyTorch, Transformers, and model adaptation techniques.…

Xiao Wang

I’m Xiao Wang, an MPhil researcher in Computer Science at the University of Sydney, focused on medical AI for histopathology vision-language understanding. My first-author work has been published at ECAI, where I investigated how vision-language foundation models can better perform zero-shot tumour malignancy recognition through retrieval-augmented and retrieval-denoising language modeling approaches. I’m also building next steps in my research, including instruction-guided retrieval refinement and knowledge-preserving domain adaptation so models can learn new domains without losing their zero-shot and retrieval capabilities. Beyond research, I’ve contributed as an AI engineer and algorithm engineer—working with real-world sensing and computer vision pipelines—while continuously strengthening my engineering toolkit with PyTorch, Transformers, and model adaptation techniques.…

Available to hire

I’m Xiao Wang, an MPhil researcher in Computer Science at the University of Sydney, focused on medical AI for histopathology vision-language understanding. My first-author work has been published at ECAI, where I investigated how vision-language foundation models can better perform zero-shot tumour malignancy recognition through retrieval-augmented and retrieval-denoising language modeling approaches.

I’m also building next steps in my research, including instruction-guided retrieval refinement and knowledge-preserving domain adaptation so models can learn new domains without losing their zero-shot and retrieval capabilities. Beyond research, I’ve contributed as an AI engineer and algorithm engineer—working with real-world sensing and computer vision pipelines—while continuously strengthening my engineering toolkit with PyTorch, Transformers, and model adaptation techniques.

See more

Work Experience

AI Engineer, AI Research and Development Department at SynerGlobal Co., Ltd.
December 1, 2025 - August 1, 2026
Conducted research on elderly fall detection using chest-worn 6-axis IMU data. Built preprocessing pipelines for accelerometer/gyroscope signals, constructed event windows, and developed fall/non-fall activity modelling. Explored domain adaptation for wearable sensing by pretraining IMU representation encoders on public fall-detection datasets and fine-tuning on company-collected elderly-care sensor data. Developed and evaluated an event-centred dual-stream accelerometer/gyroscope model using self-supervised pretraining, supervised contrastive learning, and phase-aware multi-task learning across pre-impact, impact, and post-impact phases.
MPhil Researcher (Research Experience) at University of Sydney
October 1, 2024 - June 1, 2026
Developed Retrieval-Denoising Causal Language Modelling (RDCLM) for pathology vision-language foundation models for zero-shot tumour malignancy recognition, including an LLM-based histopathology knowledge base and a denoising language-model module to improve image–text semantic alignment. Evaluated on public pathology cancer datasets and reported improvements over the strongest competing method; led model design, implementation, data processing, experiments, ablations, and academic writing. Ongoing work includes retrieval-filtered generation for instruction-guided evidence refinement and adaptive weight-space ensembling for knowledge-preserving domain adaptation.
Algorithm Engineer at Jiangsu Titan Zhihui Co., Ltd.
December 1, 2023 - March 1, 2024
Prepared and annotated video/image datasets for classification and detection using FFmpeg, LabelMe, and OpenCV, including data augmentation pipelines. Trained and evaluated object detection models to support model selection for deployment on the company’s software platform.
Software Development Intern at Zhejiang University Binjiang Research Institute
September 1, 2023 - December 1, 2023
Applied machine learning to predict tobacco chemical components from meteorological features in Henan, China. Surveyed NILM (non-intrusive load monitoring) research to support technical planning for a load-identification project.

Education

Master of Philosophy in Computer Science at University of Sydney
October 1, 2024 - June 1, 2026
Bachelor and Master of Engineering in Computer Science at University of Bristol
September 1, 2018 - July 1, 2023

Qualifications

Add your qualifications or awards here.

Industry Experience

Healthcare, Software & Internet, Education, Professional Services, Computers & Electronics