Staff Data Scientist - Wildfire

The climate crisis is the defining challenge of our time—but it’s also the greatest opportunity for innovation, and a challenge we’re proud to take on. At Overstory, we’re harnessing cutting-edge technology to enable a resilient electrical grid that keeps communities thriving as our world changes. The grid is the backbone of life as we know it. It powers hospitals, keeps food fresh, and ensures communities stay connected.

But extreme weather, aging infrastructure, and growing wildfire risks are putting this critical system under pressure. All of this combined makes the electric utility industry the greatest opportunity for tackling climate change. One of the leading causes of catastrophic wildfires and power outages? Trees and brush coming into contact with power lines. That’s where we help. At Overstory, we use AI and advanced satellite imagery to pinpoint and prioritize vegetation risks before they materialize.

By giving utilities critical analysis on those risks, we’re helping prevent outages, reduce wildfire risks, and accelerate the transition to a safer, more resilient grid. Our team spans the Americas and Europe, and we work with utility partners across the Americas and beyond. We’re outdoor enthusiasts, musicians, artists, athletes, parents, and adventurers. What unites us is a passion for solving complex problems, a commitment to climate action, and the belief that technology should be a force for good.

Join us to help us build a more resilient world together.

We are excited to add a Staff Data Scientist, Wildfire to our team. This individual will lead the scientific foundation of our Fuel Detection Model, the core engine that translates satellite and environmental data into an understanding of vegetation structure, fuel loads, and wildfire risk. Working alongside ML engineers, and a product team, you'll define accuracy for our models, design the research and validation methods that prove it, and ensure our modeling choices are grounded in fire science and remote sensing fundamentals.

This is a great opportunity for someone who is energized by open scientific questions with direct real-world stakes, and who wants their research to shape how utilities prevent catastrophic wildfires.

Time Zone Requirement:
North America (NST, AST, EST, CST, MST, PST)

• Lead research into how vegetation structure, fuel conditions, and wildfire risk can be estimated from satellite, LiDAR, and environmental data across diverse geographies
• Design rigorous validation and evaluation methodologies, including ground-truth strategies, uncertainty quantification, and error analysis tied to real-world impact
• Prototype and refine ML modeling approaches, then partner with ML engineers to translate them into production systems
• Integrate established fire science, such as fuel models and fire behavior frameworks, with data-driven methods
• Define scientific standards for experimentation, reproducibility, and model interpretability across the modeling organization
• Communicate research findings clearly to engineers, product teams, customers, and the broader wildfire science community
• Mentor ML engineers on scientific methodology and domain reasoning

• 8+ years of applied research or data science experience in wildfire science, fire ecology, forestry, remote sensing, atmospheric science, or a related quantitative field
• Deep expertise in remote sensing and geospatial analysis, including working with satellite imagery and large-scale environmental datasets
• Strong statistical modeling and machine learning skills in Python, with tools like Geo Pandas, scikit-learn, PyTorch, or XGBoost
• Track record of designing validation studies and evaluation frameworks for environmental or geospatial models
• Excellent communication skills, with the ability to make complex scientific work legible across technical and non-technical audience
• Nice To Have Familiarity with fire behavior or fuels modeling frameworks (e.g., Rothermel-based models, LANDFIRE fuel classifications)
• Experience integrating physics-based models with ML, or with active learning and uncertainty quantification
• Peer-reviewed publications in wildfire…