Senior Climate Machine Learning Scientist

About the job

• Build physical climate risk models using state-of-the-art methodologies including AI numerical weather prediction applied to climate applications; stochastic ensembles; earth observing foundation models; as well as traditional physics-based models.
• Collaborate with our climate scientists, hydrologists, fire scientists, data scientists, and data engineers to translate business requirements into scientific solutions.
• Partner with data, cloud, and security teams to design scalable and maintainable scientific modeling pipelines.
• Support a culture of scientific rigor and excellence in model design, including conducting validation studies to guide continuous improvement of those models.
• Architect and implement machine learning solutions for climate risk assessments.
• Produce robust, reproducible models which will stand up to validation studies in a changing climate.
• Follow team standards for machine learning applications to create physical risk models.
• Keep up with the latest developments in machine learning, extreme weather, and climate and bring those developments to improve our models and risk assessments.
• Train machine learning models written in PyTorch or JAX, then extend them for large scale inference on global scales.
• Use dashboards such as mlflow, Weights and Biases, etc. to track and communicate model performance and results.
• Be familiar with the latest foundation models such as Aurora, and the latest prediction models such as FourCastNet 3.
• Ph.D. in Meteorology, Atmospheric Sciences, Oceanography, Hydrology, Physics, Geography, Engineering, Climate, Mathematics, or similar subjects.
• 2+ years of experience in roles developing products to meet client requirements.
• Programming experience in Python, R, C/C++, and/or Fortran.
• Experience with cloud optimized geospatial formats such as Cloud Optimized GeoTIFF (COG), GeoParquet, and related standards.
• Knowledge of the Coupled Model Intercomparison Project (CMIP) outputs and its shared socioeconomic pathways.
• Practical experience using AI tools (i.e. Claude Code, Codex) for writing production scientific code and pipelines.