In Models Research & Development, I’ve worked on projects to make modelling processes more efficient and easier for analysts to use. I helped simplify and upgrade the main Python modelling repo in AzureML, halving average model turnaround time from 24 months to 12.
I built a PowerBI Model Monitoring Dashboard used by various teams to track model performance monthly and quarterly, identify areas for improvement, and target models for refresh.
I developed a DataBricks pipeline to process customer data from brokers at quote time, with a focus on quick, optimised calculations under timeout restrictions. I also created an underwriting adjustments dashboard, first in R and later ported and upgraded in Python.
My work has also included combining GLMs and GBMs in layered models, and building a Folium module to view geographic model effects on interactive maps. I’m guided by the idea that tedious tasks should be automated and tasks requiring human input should be simple.
