Projects
This is the public part of my work, most of it from my years at CIAT. My industry work for Globant’s clients isn’t public, so it appears only as a summary on the CV.
Forecasting the El Niño signal
The Oceanic Niño Index (NOAA Climate Prediction Center data) through September 2026, the El Niño signal behind much of the year-to-year swing in Colombia’s rainfall, with a 24-month forecast and its 50% and 90% intervals. A state-space model fitted with a Kalman filter, in about 200 lines of Python.
AClimate Colombia
A system that delivers seasonal climate forecasts and agro-climatic information to support climate risk decisions in Colombia. I worked on the statistical models behind it as part of the team at CIAT.
Seasonal forecasts for rainfed maize
Applying agro-climatic seasonal forecasts to rainfed maize management in Colombia. My R code for USAID maize forecasting at CIAT is public, with 126 commits.
Agro-climatic hotspots (AgMetGaps)
Where in the world could climate services help close crop yield gaps? I worked on the data processing and on the statistical and machine learning models.
Climate change and crops in Latin America and the Caribbean
Simulations of near-term climate change impacts on major crops, built on crop models and statistical analysis.
Climate vulnerability of Espeletia in the páramo
An assessment of how climate change threatens the Espeletia complex on the páramo sky islands of the northern Andes. Co-author.
The spectrum of cumbia
An undergraduate thesis in statistics at Universidad del Valle that I co-directed. It extracts fundamental-frequency time series from 30 cumbia songs in R and summarises them with principal component analysis. In Spanish.