Teaching
CLMT 5043: Machine Learning for Climate Science and Environmental Sustainability
Machine Learning for Climate Science and Environmental Sustainability introduces students to machine learning methods through problems in climate science, mitigation, adaptation, and environmental decision-making. The course emphasizes physically grounded problem formulation, critical evaluation, uncertainty, and the policy implications and environmental costs of AI.
I teach the course in Columbia's Climate School, where it was offered in Summer 2025 and Summer 2026. Course materials for these classes are available as an online Jupyterbook:
Online textbook (Summer 2026) ·
accessible version
Online textbook (Summer 2025)
ML in Python for Environmental Science Problems
I organized and taught this introductory short course at the American Meteorological Society Annual Meeting, for researchers interested in applying machine learning and data science to environmental research questions. Participants worked with real-world data and built ML pipelines in Python using Jupyter notebooks. A beginner section covered supervised and unsupervised learning and an introduction to deep learning, walking through the full pipeline from exploring a dataset and its features, to training models on tabular and image data, to evaluating and comparing models. An intermediate section covered physics-informed and Transformer-based architectures, along with eXplainable AI (XAI) techniques for investigating what a trained model has learned. Course materials are available on GitHub: