Jorge Cardona

I conducted research in stochastic analysis, focusing on heterogeneous diffusion, Bessel processes, and the time reversal of diffusions. My objective was to classify the weak solutions to heterogeneous diffusion and unveil its connection to skew-Bessel processes and issues of non-uniqueness.

I conducted research on machine learning and natural language processing like question answering, fact checking, and text summarization. I worked in understanding fact-checking systems and its limitations with fallacies and reasoning, as well as proposing and implementing modifications of some transformer-based architectures like BERT, T5 and BART, in order to improve the performance across various tasks.

I conducted research on problems of non-uniqueness and selection methods applied to random dynamical systems and equations of fluid dynamics. I taught a course on stochastic partial differential equations and assisted in functional analysis courses. I published a paper on the application of a measurable selection to random dynamical systems and rough differential equations.

I develop and optimized the backend and frontend components of a civil engineering collaboration software. I used Python and Javascript in the backend to construct an API in Django, Celery and some tasks running on Node.js. For the frontend, React.js and React Native with Couchbase and Couchbase Lite for an offline-first support. I worked extensively with AWS to improve scalability and reliability, and I helped to enhance the API performance with some careful analysis of the database usage.