Contributing to the development of AIO, a research infrastructure designed to support the ethical collection of digital platform data for social research. Working on platform workflows, researcher-facing tools and documentation to support study configuration, participant engagement and data donation.
Developing a researcher- and participant-oriented framework to simplify digital data donation for social research. The work focuses on helping researchers identify relevant data components, explore sample platform exports, and design accessible donation workflows while addressing technical, ethical and privacy considerations. development of a taxonomy for standardising access to digital platform data for social research. Categorising data components across platforms, including content, interactions and engagement, user information, advertisements and preferences, and metadata, to support consistent identification and interpretation of platform data.

Contributing to the STEM Indigenous Impact Project Grant, in collaboration with RMIT’s Office of Indigenous Education, Research and Engagement. Analysing Student Experience Survey (SES) data to generate evidence-informed insights into student experiences and reconciliation at RMIT. I analysed five years (2020 to 2024) of RMIT’s SES responses to questions about reconciliation. The findings were shared at a workshop with 30 university stakeholders and leaders, which looked at how SES data can be used to evaluate Responsible Practice over the next three years. The work aims to improve understanding of how Indigenous and non-Indigenous students see responsible practice, cultural awareness, and cultural safety. Findings were presented at the AARE 2025 Conference in Newcastle, and the work has led to a research article on education in Australia, which is currently under review.

Investigating challenges in forecasting time series data from diverse sources. Developing deep learning architectures to model irregular, incomplete, and heterogeneous time series, including high-dimensional streaming data and online learning scenarios. This work was part of my PhD thesis and has led to several publications, including papers in Transactions on Machine Learning Research (TMLR) and the IEEE International Conference on Data Mining (ICDM). The models are based on neural ordinary differential equations (ODEs) and parallelised encoders.

Developing and documenting an audio-visual Arabic emotion dataset comprising 3,000 video and audio clips across six basic emotion categories. Designing and evaluating deep learning models for emotion detection and recognition using audio-only, video-only and multimodal approaches.