Experience

March 2022 - Present Postdoctoral Fellow in Machine Learning at CSIRO

– Developed a Bayesian hierarchical model for linear regression between metal grades for uncertainty evaluation and reduced the posterior predictive variance of the metal grade
by up to 59.14% to aid stakeholders for decision-making.
– Developed a Bayesian linear regression with Gaussian Mixture Likelihood for outlier detection in the linear regression between metal grades and reduced the mean square error up
to 99.5% compared with maximum likelihood estimation, supported decision-making for stakeholders.
– Designed a fine-tuning solution based on GAN for digital image denoising from large computer vision model – VGG19 and improved perceptual image quality (SSIM) by 8.78%.
– Developed advanced deep learning models (Transformer, LSTM, RNN) and statistical models (ARIMA) on time series data prediction from physical sensors for decision-making in
sorting.
– Designed advanced deep learning models (Self-supervised learning: Noise2Void, Noise2Noise; Generative AI models: Diffusion related models, GAN based models; Supervised methods:
CNN based models) to denoise digital ore images and dramatically improved the speed of image reconstruction by 75%.
– Utilised Power BI to visualise data for stakeholders.
– Supported various colleagues on designing advanced deep learning algorithms (ControlNet, Diffusion, Transformer, GAN, CNN).
– Presented on internal/external seminars and conferences.
– Wrote/Published scientific journal papers.
– Supervised a Ph.D. student at UNSW on physical-guided machine learning for spectrum data analysis

Jan 2022 - March 2022 Machine Learning Specialist at UNSW

– Designed artificial neural networks for physical property prediction
– Implemented deep learning models for material characterization
– Conducted research on neural network architectures for predicting material behaviour
– Utilised machine learning algorithms to forecast physical properties
– Developed AI models to simulate material responses