London needs to know its future population to plan for school places, housing and infrastructure. I’ve been building the next generation of population models to do this.

• Creating a modular demographic model that enables customisation and experimentation.
• Introducing stochastic uncertainty to the model based on timeseries analysis of migration data.
• Exploring how machine learning and new data sources can improve the modelling of birth and death rates
• Implementing version control, Agile methodolgies, coding best practices and code review for the team. Making sure our published results are fully reproducible from the raw data.
• Liaising with stakeholders and academics to better guide model design
• Work in R, bash, SQLite, SPARQL
  

I worked on the Event Response team, writing and maintaining code that to forecast and reconstruct natural disasters in near-real time. I specialised in European extratropical cyclones and western Pacific tropical storms, but also worked to model floods, wildfires, earthquakes, severe convective storms and hurricanes.

• Forecasting and reconstructing natural disasters using observational data, weather agency forecasts and the RMS models.
• Improving, extending and automating code. Adapting existing processes to work with new models and regions.
• Designing and building new products.
• Writing daily briefings and reports for clients.
• Work in R, bash, PostgreSQL, with some perl, python and Fortran.
  

I worked with the European Flood team, adapting the in-house flood model to include the effects of climate change. My work took an event set of 50,000 years of stochastic rainfall data and came up with a novel reweighting method to be consistent with different IPCC climate change scenarios. The work is now being published.

My team was funded by Manchester University Press and the University of Manchester’s REALab scheme to conduct a short research project. With interviews and research we produced an in-depth review of the changing landscape of digital academia for the discovery phase of a new, digital platform for the Press.

My research quantifed the range of structures found in intense small-scale Arctic storms - polar lows. I used a cyclone tracking routine to identify polar lows in historical reanalysis data, and automated the analysis of their structures, adapting a technique used to study the transition of hurricanes into midlatitude storms. This let me identify storm structures associated with the most hazardous weather. Funded by the Natural Environment Research Council.

• An extended, self-directed project.
• Analysing very large datasets with NCL, with some work in C and Fortran.
• Running case study simulations and sensitivity tests with the WRF and Polar WRF models, which I compiled, optimised and ran on the UK’s ARCHER HPC supercomputer.
• Presenting work internationally and publishing results.