Projects

Selected Foundational Research Projects:

  • Continual Self-Improving ML | London, UK (2025–Present). My primary research focus explores integrating self-improvement capabilities into continual learning systems via transfer mechanisms and meta-learning. Active Status: Currently seeking funding and research collaborators.

  • Knowledge Transfer in Continual Learning | Auckland, NZ (2021–2025). Investigated mechanisms to facilitate knowledge transfer while mitigating catastrophic forgetting in supervised continual learning. Role: Principal Investigator. Funding: Marsden Fast Start Grant (Royal Society of NZ), New Staff Research Grant (University of Auckland). Collaborators: Bing Liu (UIC), Patricia Riddle (UoA).

  • Long-Sequence Transformer Architecture | Auckland, NZ (2021–2024). Developed novel transformer architectures optimized for long-sequence data. Led the research team in architectural exploration, experimental design, and performance evaluation. Collaborators: Michael Witbrock, Joshua Bensemann, Alex Peng, Trung Nguyen, Neset Tan (UoA).

  • Deep Learning for GWAS | Auckland, NZ (2021–2023). Applied state-of-the-art transformer architectures to genomic data to predict gout susceptibility using public and NZ-specific datasets. Led the computer science research stream. Collaborators: Michael Witbrock, Trung Nguyen, Neset Tan (UoA); Alex Gavryushkin, Kieran Elmes (Univ. of Canterbury).

  • A Max-Margin Continual Learner | Auckland, NZ (2021–2022). Adapted continual learning methods from Support Vector Machines (SVM) to deep neural networks. Role: Principal Investigator. Collaborators: Patricia Riddle, Gillian Dobbie (UoA).

  • Human-Centered Machine Learning | Auckland, NZ (2020–2021). Examined the challenges of human-algorithm collaboration, specifically focusing on how ML models support and influence human decision-making processes. Collaborators: Rhema Vaithianathan (AUT), Gayani Tenakooon (UQ).

  • A Framework for Long-Term Learning Systems | Auckland, NZ (2016–2019). Investigated methodologies for sequential task learning using SVM frameworks. Role: Principal Investigator. Collaborators: Patricia Riddle, Yun Sing Koh (UoA).

Selected Applied Research Projects:

  • Predictive Risk Models for Child Maltreatment (2026–Present). Researching methodological and technical enhancements for deployed ML tools to improve the accuracy and ethics of child maltreatment referral decision

  • Decision Aid Tool | Los Angeles County, CA, US (2020–2021). Led the Data Science and Software Engineering teams in constructing and deploying an ML tool to screen child maltreatment referrals. Collaborators: Rhema Vaithianathan (AUT), Emily Putnam-Hornstein (USC), Alexandra Chouldechova (CMU). Partners: LA Dept. of Children and Family Services, Analytics for Change. Methodology Report.

  • Hello Baby Prevention Program | Allegheny County, PA, US (2019–2020). Spearheaded the development of a proactive ML tool to identify children at high risk of maltreatment for early intervention programs. Collaborators: Rhema Vaithianathan (AUT), Emily Putnam-Hornstein (USC), Alexandra Chouldechova (CMU). Partner: Allegheny County Dept. of Human Services. Methodology Report.

  • Decision Aid Tool | Douglas & Larimer Counties, CO, US (2018–2021). Directed the end-to-end development of screening tools, including rigorous evaluations of model fairness, subgroup discrimination, and interpretability. Collaborators: Rhema Vaithianathan (AUT). Partners: Douglas/Larimer County Dept. of Human Services, Mathematica Public Policy. Methodology Report.

  • Allegheny Family Screening Tool | Allegheny County, PA, US (2016–2021). Managed the engineering and data science lifecycles for a decision-support tool, with a specific focus on algorithmic fairness and bias mitigation. Collaborators: Rhema Vaithianathan (AUT), Emily Putnam-Hornstein (USC), Alexandra Chouldechova (CMU). Partner: Allegheny County Dept. of Human Services. Project Details.

  • Prototype Model | Alerta Niñez, Santiago, Chile (2018–2019). Identified and evaluated ML architectures for a nationwide child protection initiative, leading the assessment of model fairness and social impact. Collaborators: Rhema Vaithianathan (AUT), Maria Hermosilla (AIU). Partner: Ministry of Social Development, Chile. here.