CV

Download my CV (PDF)

VIC, Australia · mailto.xiaoyu.guo@gmail.com · Google Scholar · LinkedIn

Research Profile

AI researcher with a PhD in Artificial Intelligence from Monash University and research experience spanning academia and industry. My research focuses on natural language processing, large language models, complex reasoning, multimodal learning, and responsible AI, with publications in leading international conferences, including EMNLP and COLING. I have a strong track record of developing novel machine learning methods, constructing evaluation benchmarks, and investigating model robustness, fairness, and generalisation. My experience includes conducting independent and collaborative research, supervising research students, and translating research outcomes into practical AI applications. I am particularly interested in advancing trustworthy, efficient, and generalisable AI systems.

Professional Experience

Senior Applied Scientist · Oracle Corporation Australia Pty Limited

January 2025 – May 2026 · Melbourne, VIC

Applied science role spanning real-world data handling, ambiguous problem solving, medical-domain agent workflows, model evaluation and validation in a customer-facing environment.

• Real-world data handling: Consolidated domain knowledge from customers and subject-matter experts, and built automated pipelines for real-world data selection, ingestion, preprocessing, and post-processing to support downstream AI workflows.
• Ambiguous problem solving: Worked closely with stakeholders to clarify ambiguous requirements and designed scalable AI solutions that translated evolving requirements into deployable systems.
• Agent workflow design: Designed and implemented agent workflows within the medical domain, integrating LLM-based automation to improve task execution and usability.
• Model evaluation and validation: Conducted evaluation experiments, including performance testing, validation, and robustness assessment of deployed models to ensure reliability and effectiveness in real-world settings.

Postdoctoral Research Fellow · Australian Institute for Machine Learning

November 2023 – January 2025 · Adelaide, SA

Research role focused on machine learning, system robustness, responsible AI, algorithm design, model optimization, technical documentation, and publication outputs.

• Machine learning research: Conducted research on fundamental machine learning challenges, with a focus on robustness, reliability, and responsible AI.
• Algorithm design and evaluation: Analyzed system requirements, designed novel algorithms, and evaluated their effectiveness through empirical experimentation using quantitative metrics.
• Model optimization: Collaborated with colleagues to refine technical specifications, investigate system limitations, and improve model performance.
• Research communication: Contributed to technical documentation and research publications, and presented findings, methodologies, and results to the research community.

Education & Credentials

Doctor of Philosophy, Artificial Intelligence · Monash University

March 2020 – April 2024 · Faculty of Information Technology · Melbourne, VIC

Full scholarship recipient.

Thesis: Complex Question Answering: from Data to Method

Research focus: Designed and implemented machine learning models, including Neural Module Networks and LLM-based single- and multimodal architectures, to address complex question answering tasks under real-world data bias and distribution challenges.

Master of Engineering, Computer Science and Technology · Beihang University

September 2016 – January 2019 · School of Computer Science and Engineering · Haidian, Beijing

Honors & awards: Outstanding Postgraduate Student in Beijing (2019); Outstanding Postgraduate Student, Beihang University (2019).

Bachelor of Engineering · Beihang University

September 2012 – June 2016 · School of Computer Science and Engineering · Haidian, Beijing

Selected Publications

First Author

Question answering, natural language processing, and relation classification.

• DeSIQ: Towards an Unbiased, Challenging Benchmark for Social Intelligence Understanding. EMNLP 2023.
• Complex Reading Comprehension Through Question Decomposition. ALTA 2022.
• Improving Numerical Reasoning Skills in the Modular Approach for Complex Question Answering on Text. Findings of EMNLP 2021.
• Understanding Unnatural Questions Improves Reasoning over Text. COLING 2020.
• Automated Thematic and Emotional Modern Chinese Poetry Composition. NLPCC 2019.
• A Single Attention-Based Combination of CNN and RNN for Relation Classification. IEEE Access, 2019.

Coauthored Papers

• Multi-hypothesis Test-Time Adaptation to Mitigate Underspecification. ECCV 2026.
• Bayesian Low-Rank Learning (Bella): A Practical Approach to Bayesian Neural Networks. AAAI 2025.
• Mitigating Language Bias of LMMs in Social Intelligence Understanding with Virtual Counterfactual Calibration. EMNLP 2024.
• An Empirical Analysis on Spatial Reasoning Capabilities of Large Multimodal Models. EMNLP 2024.
• Teaching Neural Module Networks to Do Arithmetic. COLING 2022.
• Compose Like Humans: Jointly Improving the Coherence and Novelty for Modern Chinese Poetry Generation. IJCNN 2020.

View the complete publication list and citations.

Selected Earlier Research Experience

Research Assistant on Computational Cultural Understanding · Monash University

February 2022 – July 2023 · Faculty of Information Technology · Melbourne, VIC

• Norm detection and discovery: Without labeled data, built a model to classify social norms such as greetings and apologies in Chinese culture, using multimodal inputs including videos, audio, and transcripts.
• Social intelligence question answering: Based on the Social-IQ benchmarks, focused on fusion and reasoning over multimodal data, proposed a methodology for dataset bias detection and debiasing, and had a paper accepted at EMNLP 2023.

Internship on Chinese Poetry Generation · JD AI

July 2018 – May 2019 · NLG Group · Chaoyang, Beijing

• Modern Chinese poetry generation: Started from a basic seq2seq model with attention, incorporated human emotions and topics, and conducted experiments on millions of data points to achieve more coherent, expressive, and thematic poem generation.
• Modern Chinese poetry refinement: Developed a follow-up generate-then-refine approach, adding encoders for useful poem lines.
• Knowledge-driven dialogue system: Leveraged knowledge graphs and OpenNMT to generate multi-turn dialogue between machines and humans.

Research Assistant on Relation Classification · Beihang University

March 2018 – January 2019 · State Key Laboratory of SDE · Haidian, Beijing

• Relation classification: Built a novel model combining RNN and CNN, conducted experiments on the SemEval-2010 Task 8 dataset, and achieved the highest F1 score compared with previous state-of-the-art models.
• Refined relation classification model: Extended the model and datasets to SemEval-2018 Task 7 and KBP37, and published a paper in IEEE Access.

Scholarships & Honors

• January 2019: Outstanding Postgraduate Student in Beijing.
• March 2018: Outstanding Postgraduate Student in Beihang University (20 out of 235).
• October 2017: Second Prize Scholarship in Beihang University.
• June 2016: Outstanding Undergraduate Student in Beihang University.
• October 2015: Outstanding Student Leader in Beihang University.
• December 2014: Learning Outstanding Scholarship in Beihang University.

Competition Awards

• May 2015: First Prize in the Asia Student Supercomputer Challenge.
• November 2014: CCF score: 300; ranking: 5.73%.

Association Participation & Volunteering

• September 2014 – July 2015: SCSE 120615 class monitor at Beihang University.
• July 2014: INCLUDED volunteer (July 14–25).