Parthib Paul

I worked with the Risk team automating how the company tracks and prioritizes risk. Built a 3-layer data model in Sigma that's been used by the Accounting team for capital ratio calculations. I also designed their entire data-intake workflows and migrated their entire Enterprise Risk Register from Excel into centralized dashboards so that everyone could evaluate risk the same way instead of doing their own thing in spreadsheets.

I joined the Language Information and Computation Lab and worked on making large language models less overconfident when you distill them into smaller ones. Tested a bunch of model configurations and cut the calibration error by more than half without hurting output quality.

I worked on modeling how forests and wildlife recover after wildfires. Built Python scripts that combined field ecology data with GIS to track post-fire growth of 220+ trees across Waldo Canyon burn sites and used regression models to figure out how burn severity actually affects wildlife nesting and vegetation recovery.

I built and maintained a full-stack Django app that scaled to potential 2k+ daily users of my college. Handled everything from backend features and REST APIs to containerized deployments with Docker and CI/CD.

I researched how accessible makerspaces help disabled students build skills and career readiness. Ran interviews across 3 sites, transcribed audiovisual data in Descript, and built 11-code frameworks in Dedoose. The work got accepted and presented at The 27th International ACM SIGACCESS Conference on Computers and Accessibility 2025.