Machine Learning Wildfire Scientist

Coulson Aviation • Thermal, CA

12d • Remote

About The Position

EmberWorks Sensing, the remote sensing arm of Coulson Aviation’s R&D division, operates ORBIT, a persistent aerial thermal sensing system flown on Coulson air attack aircraft. ORBIT captures and delivers near-real-time fire intelligence to tools used by incident teams.

The Machine Learning Wildfire Scientist will own EmberWorks’ fire behavior modeling effort, from integrating existing models to developing new methods. This is a founding position and a new function for the company. The scientist will shape the modeling approaches, validation methods, and delivery of model outputs to end users. They will work closely with a small team of geospatial and platform engineers responsible for the data pipeline and cloud infrastructure, and serve as the team’s modeling expert.

The position is remote. The work focuses on using airborne thermal observations, together with external fuel, weather, and terrain data, to build and validate fire behavior models. It includes selecting and implementing suitable approaches, developing empirical, physical, or machine learning methods, establishing model validation and interpretation practices, defining computing needs for training and inference, and engaging with the academic fire science community and research partners.

Requirements

• Deep understanding of fire behavior models, including their underlying physics and mathematics, not only experience running existing models.
• Hands-on experience building and training machine learning or data-driven models on spatiotemporal data.
• Strong remote sensing background, including experience working with raster imagery.
• Proficiency with a scientific geospatial programming stack, such as Python xarray, GDAL, rasterio, or PyTorch, or similar tools.
• Familiarity with Git-based version control.
• Experience running model training on cloud compute, including GPU resources.
• Ability to work independently in a small R&D team and explain modeling decisions to engineers and operations staff.
• Bachelor’s degree in fire science, atmospheric science, remote sensing, computer science, environmental science, or a related field.

Nice To Haves

• Thermal remote sensing experience.
• Experience delivering at least one wildfire project end to end, or sustained, in-depth support of an established wildfire product suite.
• Active ties to academic or government fire research groups.
• Experience with data assimilation or physics-informed machine learning.
• Knowledge of optics and radiometry, including thermal sensor calibration.
• Familiarity with wildland fire operations and the agencies and tools that support them.
• An advanced degree and research in fire science, atmospheric science, remote sensing, computer science, environmental science, or a related field, or equivalent applied experience.

Responsibilities

• Evaluate existing fire behavior modeling approaches and select approaches suited to EmberWorks’ data.
• Build and implement models in an automated production pipeline using airborne thermal observations and external fuel, weather, and terrain data.
• Research and develop modeling approaches using empirical, physical, and machine learning methods as appropriate.
• Own model validation and interpretation by defining quality criteria, detecting bad outputs, and distinguishing model problems from upstream data problems.
• Define compute requirements for training and inference and collaborate with the platform engineer on GPU and distributed computing resources and an efficient remote workflow.
• Maintain connections with the academic fire science community and help EmberWorks engage with research partners.

What This Job Offers

• Job Type: Full-time
• Career Level: Senior