Inference in privacy-preserving data processing systems

Within this Horizon project we study Bayesian data augmentation methods that can ease decentralized inferential tasks with incomplete data.

The work is motivated by privacy and security concerns raised by individuals, governments and organizations currently relying on solutions designed for shared databases of personal information, and brings together an interdisciplinary team of computer scientists and mathematicians.


healthTechnologies to inform best practice in secondary care

I was part of Wayward, an interdisciplinary project investigating the use of new technologies in order to drive improvements in safety and efficiency within secondary healthcare settings. The team was partnered with NHS Trusts, the Liverpool School of Tropical Medicine and the Royal College of Physicians.

My contributions analyzed a variety of temporally-ordered data and meta-data, aiming to study dependence structures among variables and draw inference in relation to clinical behaviour and workload.

ballSports modelling

As part of the quantitative team at Sportradar, I have developed predictive models for competitive sports such as tennis, basketball and Australian rules football.

On one hand, I have worked on signal processing and non-linear filtering methods aiming to track the performance and abilities of players over time. Also, I have studied copula approaches and simulation-based methods allowing to approximate multivariate distributions of counts for multiple correlated sporting events.

stockStochastic processes and applications to finance

My doctoral research is focused on optimal stopping and wealth allocation problems in finance. First, my thesis links to the study of optimal times for taking particular actions; such as selling or buying goods when facing uncertainty in relation to future prices.

In addition, the work is concerned with analyzing wealth allocation techniques within defaultable financial markets, including loans or bonds. In both cases, we aim to derive decision rules in order optimize rewards.