Bio
Viv has had a lifelong affinity for Mathematics, consistently demonstrated through outstanding academic performance. This paved the way for a fully funded PhD position in Mathematics and Statistics at the University of Exeter through the Engineering and Physical Sciences Research Council (EPSRC). Prior to starting her PhD, she graduated top of her class with a Bachelor of Technology degree, earning double honours in Physics and Computer Science. Viv has enjoyed a versatile career in financial services organizations and technology companies, with roles spanning programming, technical sales, and product management. This broad skill set and adaptability reflect her diverse experience and passion for research, which ultimately led her back to academia. Her Master’s thesis explored the use of Neural Networks for post-processing weather forecasts, and she is now expanding her interest in statistics and data science through her PhD project, which aims to integrate deep learning methods with statistical spatiotemporal models. Outside of her professional and academic pursuits, Viv enjoys writing and has co-authored and published computer science textbooks for primary and secondary schools. Additionally, she is an avid runner and yoga practitioner, finding balance and inspiration through her active lifestyle.
Research interests
Viv’s research is funded through the Engineering and Physical Sciences Research Council (EPSRC) and the research goal is to study space-time models motivated through physics-based stochastic partial differential equations (PDE), to solve complex environmental problems. One application area is the problem of precipitation nowcasting, which is crucial in providing early warning of extreme weather events. This method is inspired by the physical advection - diffusion PDE, which can be written as a vector autoregressive VAR(1) process, leveraging sparse matrix methods and Laplace approximations to carry out Bayesian inference on large amounts of data common in spatiotemporal applications. For operational use in next generation numerical weather prediction, she aims to extend this method using deep learning techniques, for example Physics Informed Neural Networks (PINNs). This work ties in with Turing core capabilities foundational AI key focus area of programmable inference for large PDEs. She also hope to show the potential generalization and transferability to other application areas like epidemiology and economics where the ability to provide timely estimates of the present and immediate future remain desirable. For example, in epidemiology, nowcasting techniques may be applied to death counts for correct estimation of number of deaths due to an epidemic, improving our understanding of the disease and effective public health planning. The key aspects of simulation, estimation and prediction as applicable in epidemiology or economics, fits with both AI for public services and Machine Learning in public health interventions projects.