My work sits between mathematical structure and practical machine-learning problems.

I began in spectral graph theory and magnetic Laplacians, and my recent work applies that structural perspective to NLP, graph representation learning, scalable clustering, and graph-signal analysis for scientific data.

The common thread is structured complexity: data where relationships matter as much as individual observations.

Across research appointments in Cardiff and Edinburgh, I have built Python and PyTorch workflows, designed measurable evaluation pipelines, and turned open-ended technical questions into reproducible prototypes and comparative studies.

Earlier in my career, I worked in a government data-science setting in Mexico, where I led modelling work on national assessment and census data, created Shiny dashboards for policy teams, and helped reduce analysis-to-decision cycles from weeks to hours.