Conducted data preprocessing of 3D micro-CT rock images, including data augmentation and image features extraction from over 13,000 images with Python and OpenCV.

Built regression models in Python and containerized the models in docker on AWS to predict permeability of rock samples, which achieved a 80% accuracy rate by MAPE.

Integrated all trading and financing positions, risk factors in respect of market risk, credit risk and operational risk at group level to implement enterprise-wide risk management

Evaluated the expected credit loss and impairments for margin financing business under IFRS 9

Developed IFRS9 models to forecast expected credit loss for financial assets with around $6B risk exposure. Was in charge of code development by R, firmwide framework implementation and risk reporting.