How to apply for Staff Machine Learning Engineer - Wildfire

About Overstory

Overstory uses real-time satellite data and machine learning to map vegetation and reduce wildfire risk, directly improving grid reliability and combating climate change. Working here means your ML models have a tangible environmental impact, helping utilities prevent wildfires before they start. The company sits at the intersection of remote sensing, geospatial analytics, and climate tech, offering a mission-driven environment for engineers passionate about sustainability.

About the role

As a Staff Machine Learning Engineer on the Wildfire team, you will own the Wildfire Fuel Detection Model, the core engine that turns satellite imagery into actionable vegetation and fuel load insights. You'll architect models that generalize across diverse geographies, build scalable data pipelines for petabyte-scale geospatial data, and partner with wildfire scientists to define metrics tied to real-world risk reduction. This is a high-impact role where your technical decisions directly influence how utilities mitigate wildfire threats.

A typical day

Your day might start with a stand-up with the Wildfire team, then dive into debugging a feature pipeline that processes terabytes of satellite imagery. Later, you'll pair with a wildfire scientist to refine evaluation metrics, then prototype a new vision transformer architecture for fuel load segmentation. You'll also review code, mentor junior engineers, and plan how to scale the model to a new region with different vegetation patterns.

Who Overstory is looking for

• 10+ years (or 6+ for exceptional candidates) building production ML systems, with deep expertise in deep learning, computer vision, or remote sensing applied to geospatial data.
• Proven track record designing end-to-end ML pipelines—from ingesting satellite imagery and environmental data to deploying and monitoring models at scale.
• Hands-on proficiency with PyTorch or TensorFlow for model development, plus tools like Dask, Spark, or GeoPandas for large-scale geospatial data processing.
• Experience collaborating with domain experts (e.g., wildfire scientists, ecologists) to translate scientific requirements into robust ML solutions and evaluation metrics.

Tips for this application

• Highlight specific projects where you worked with satellite imagery, aerial LiDAR, or other remote sensing data—quantify the scale (e.g., terabytes of imagery, global coverage) and impact.
• Emphasize production ML system design: describe how you handled data versioning, model monitoring, and retraining pipelines for geospatial or temporal data.
• Mention any experience with wildfire, vegetation, or environmental modeling—even if academic—as it shows domain alignment with Overstory's mission.
• Show familiarity with geospatial libraries (GeoPandas, Rasterio, GDAL) and distributed computing (Dask, Spark) in your resume bullets, not just a skills list.
• Tailor your outreach: if you have contacts in climate tech or remote sensing, ask for a referral; Overstory is a mission-driven startup where cultural fit and passion for the problem matter.

What to cover in your cover letter

['Your hands-on experience building computer vision or deep learning models on satellite/aerial imagery, and how you ensured they generalized across geographies.', 'A concrete example of scaling an ML pipeline from research to production, including challenges like data drift, latency, or compute constraints in geospatial contexts.', "How you've collaborated with scientists or domain experts to define evaluation metrics that reflect real-world impact (e.g., wildfire risk reduction, grid reliability).", 'Your motivation for joining Overstory specifically: connect your work to combating climate change and improving grid resilience through ML.']

Research before applying

• Explore Overstory's website and blog to understand their product offerings, especially how they use satellite data for vegetation management and wildfire risk.
• Read about the Wildfire Fuel Detection Model—look for any public talks, papers, or blog posts by Overstory engineers on their ML approach.
• Research the basics of wildfire science: fuel types, vegetation structure, and how utilities use risk assessments to prioritize mitigation.
• Check recent news about wildfires and grid reliability to understand the regulatory and operational context Overstory operates in.

Likely interview topics

Based on the job description, expect questions about:

• Design an end-to-end ML system for detecting vegetation fuel loads from multi-spectral satellite imagery, covering data ingestion, preprocessing, model architecture, and deployment.
• How would you handle class imbalance and spatial autocorrelation when predicting rare high-risk wildfire zones?
• Discuss a time you had to adapt a model trained in one geography to perform well in another with different vegetation types or atmospheric conditions.
• What evaluation metrics would you propose for a wildfire fuel detection model, and how would you validate them with domain scientists?
• Explain how you would optimize a PyTorch model for inference on large-scale geospatial tiles, considering memory and latency constraints.

Common mistakes to avoid

• Submitting a generic ML resume that doesn't highlight geospatial, remote sensing, or environmental data experience—this role requires domain-specific skills.
• Focusing only on model accuracy without discussing production concerns like data pipelines, scalability, monitoring, or explainability.
• Ignoring the mission: applying without demonstrating interest in climate tech or wildfire prevention can make you seem like a poor cultural fit.