Publications
Research · Communication · Impact
Research that advances AI for climate and human good.
My work sits at the intersection of machine learning and atmospheric science. I am passionate about making AI research accessible — through rigorous papers, public science communication, and open dialogue about AI's role in society.
Featured Publication
PiggyCast: Improving Weather Prediction Accuracy through a Stacking-Based Ensemble AI Approach
Josiah K. Kimani — AIMS South Africa & Stellenbosch University
AI Weather Prediction (AIWP) models have recently outperformed classical Numerical Weather Prediction models across multiple benchmarks. This paper introduces PiggyCast — a stacking-based ensemble that trains on forecast outputs from frontier AIWP models (hosted on WeatherBench 2) to predict ERA5 reanalysis variables. By piggybacking on the strengths of leading AI research teams, PiggyCast achieves notable improvements in Root Mean Squared Error on Geopotential Height at 500 hPa, with gains increasing at longer forecast lead times.
Conference Papers
Redesigning Assessment of Mathematics Education in the AI Era
Josiah K. Kimani
Examining how the rise of AI tools is reshaping the design and evaluation of mathematics assessments — and what educators and institutions should do in response.
Research Themes
Topics at the core of my research agenda.
Interested in Collaborating?
If our research interests overlap — AI, climate, weather prediction, or the governance of AI — I would love to hear from you.
Get in Touch →