CV

Data Scientist

Centre for Youth Mental Health, Orygen

Worked end-to-end data and machine learning pipeline including data extraction, cleaning and statistical analysis.

• Experienced in data extraction, wrangling, and statistical analysis using structured and non-structured data
• Applied machine learning algorithms and large language models in data analysis and health outcomes prediction
• Contributed to developing statistical analysis plans and writing technical reports, research papers and policy briefings
• Collaborated with clinicians, researchers, and policy makers to deliver data-driven solutions
• Developed interactive dashboards and visualisations tools for non-technical stakeholders to support data-informed decision-making

Data Processing & Machine Learning Tutor (Casual)

The University of Melbourne

Delivered tutorials and practical workshops on data processing, machine learning, and signals & systems.

• Taught key topics including data wrangling, visualisation, natural language processing, supervised and unsupervised machine learning
• Used Python libraries including Pandas, NumPy, and Scikit-Learn in teaching modern data science and advanced machine learning concepts

Doctor of Philosophy

The University of Melbourne

Statistical Signal Processing

Master of Science

University of Bristol

Image Processing

Bachelor of Engineering

Beijing University of Posts and Telecommunications

Telecommunications Engineering

Clinical NLP and Predictive Modelling in Medical Case Notes

Applied advanced natural language processing techniques and large language models (LLMs) to extract, structure, and utilise insights from unstructured clinical text. This project includes text de-identification, topic clustering and outcome prediction.

• Project 1 - Medical case note de-identification: Developed an automated, large language models (LLMs) based de-identification pipeline to identify and identify and mask personally identifiable information (PII) from clinical notes. The solution integrates external data sources (eg. location-based information) and goes beyond generic NER by incorporating real-world domain-specific knowledge.
• Project 2 - Topic Modelling and Clustering: Implemented a BERTopic-based framework to extract latent themes and group similar clinical case notes. Identified key clinical topics and themes through unsupervised clustering.
• Project 3 - Outcome prediction: Designed and validated models using structured features and text embeddings to predict clinical outcomes.

EEG Data Analysis in Music Therapy

Analysed EEG data in music therapy research, including data cleaning, preprocessing, and statistical analysis.

• EEG data analysis and visualisation
• Collaboration with health professionals, doctors and music therapists

Financial Timeseries Processing and Forecast

Performed financial data analysis and timeseries forecasting using machine learning models.

• Data cleaning and preparation: outlier detection, data visualisation and feature engineering
• Data analysis: correlation, moving-average, auto-regression analysis
• Timeseries forecasting: auto-regression model and machine learning models (decision tree, logistic regression, neural networks)

Customer Purchasing Behaviors Analysis

Analysed transaction data to find patterns in customer behavior.

• Data pipeline: cleaning, preparation and visualisation
• Draw Business insights based on analysis results

Reinforcement Learning and Multi-Armed Bandits (MABs)

Implemented and evaluated reinforcement learning algorithms including UCB and LinUCB.

• ϵ-Greedy, Upper Confidence Bound (UCB) models

The changing impacts of social determinants on youth mental health in Australia

The International Journal of Social Psychiatry

Using data linkage for mental health research in Australia

Australian & New Zealand Journal of Psychiatry

Feasibility of clinical EEG for music recognition in children aged 1–12 years

Frontiers in Pediatrics

Capturing the clinical complexity in young people presenting to primary mental health services: a data-driven approach

Epidemiology and psychiatric sciences

Correction for time-varying signal power in fNIRS connectivity analyses

Society of fNIRS Virtual Conference

Correction of induced functional connectivity in filtered resting state fNIRS data

OHBM Annual Meeting

Motion Artefact Removal in Functional Near-infrared Spectroscopy Signals Based on Robust Estimation

IEEE ICASSP

A very low complexity reduced reference video quality metric based on spatio-temporal information selection

IEEE ICIP

Girl Power Mentor

The University of Melbourne

Mentored Year 11/12 female students interested in science and engineering.

• Encouraged young women to pursue STEM education