Industry advisor and panelist for the interdisciplinary Masters program in Data Science and Artificial Intelligence

I led all the data science activities at the firm, while fostering a culture of scientific discipline and data-driven decision making. As a founding member, I established the data science capabilities at the firm from scratch, building numerous end-to-end data pipelines and providing mentorship to team members working in the analytics space. I developed an optimisation engine that is at the core of the firm's asset allocation process, trained machine learning algorithms that produced profitable trading strategies and built several interactive dashboards to visualise and explore the insights from statistical models. I also served as a member of the firm's executive team, driving the strategic direction of the business.

I taught online courses in data science, including python programming and machine learning.

As the Head of Advanced Analytics for the Momentum Metropolitan Group, I led a team of data scientists who developed and applied cutting-edge machine learning algorithms and statistical models to solve diverse problems across various businesses. The projects included object detection in images, sentiment analysis from voice data, AI-driven fund management and more traditional supervised and unsupervised learning. We worked closely with the Consumer and Market Insights team to deliver a complete analytics solution that included how to best act on the information that we provided. I also assisted senior executives and HR in understanding the role of data science and analytics in the organisation.

I was responsible for developing machine learning algorithms to extract actionable information and make predictions to inform business decisions in the health insurance space. This included the development of clinical algorithms to predict disease onset and hospitalisation risk, as well as algorithms to improve operational processes, such as automated email classification, churn modelling and fraud detection. I used a combination of R, Python and Tensorflow.Some specific projects:(1) AI-Driven Health Risk ManagementI developed machine learning algorithms to predict hospitalisation risk for high-cost diseases and identify the key factors that contribute to an individual's risk. These were combined to obtain an aggregate health risk score and profile for each member that is currently used to prioritise case management and guide the intervention strategies at MMI Health.(2) Automated Email ClassificationThe HIV contact centre is inundated with emails that are manually reviewed and forwarded to the appropriate team for processing. Using a combination of feedforward and convolutional neural networks programmed in Tensorflow, I developed a system that can automatically classify an email based on the images and text that appear in the attachments, so that it can be forwarded to the correct department without human intervention. The classifier had an accuracy of 98%.(3) Fraud DetectionI worked on a Bayesian fraud detection model to identify whether provider claims were unusual given the hospitalisation probabilities associated with the claims.(4) Predicting Fitness for DutyI trained a continuous-time Markov chain with time-varying covariates to predict whether a member of the South African police force would be fit for duty at a future time point. The model could also suggest reasons why a policeman/woman might become unfit for duty so that an intervention could be effected immediately to prevent this.

This role had research and teaching responsibilities. My research involved adapting and developing new statistical modelling and data science techniques to solve real-world problems from various fields, ranging from economics to the health sciences. This often involved collaborating with colleagues without a formal quantitative background. I published 16 peer-reviewed articles in leading journals, presented my work at several international conferences and was a nationally recognised researcher (rated by the National Research Foundation of South Africa). I taught and convened various undergraduate and postgraduate courses in statistics and data science, and designed the syllabus and material for many of these. Through this experience, I developed a deep-set passion for teaching and supervision and enjoyed thinking about how I could make difficult content accessible to my students. I led the development of the first interdisciplinary master’s degree in data science in the country, aimed at working professionals, and introduced a new undergraduate programme in business analytics. I supervised many postgraduate students, most of whom graduated with distinction. My efforts were rewarded with a Distinguished Teachers Award, the highest university accolade awarded to only a handful of academics for exceptional teaching, curriculum development and supervision.