Data Scientist

Understory; Madison (WI); On-site; USD 95,000 - 130,000; Full time

Job summary

Understory is seeking a Data Scientist to advance geospatial data, machine learning, and real-world risk insights for insurance products. You’ll work with sensor data, risk models, and a fast-moving engineering stack to deliver actionable analyses for underwriters and partners.

Responsibilities span modeling, mapping, and data wrangling, along with building internal tools and reports. Candidates should be adept at translating complex results into business impact and collaborating across teams.

Qualifications

• 3–5 years of experience in data science, catastrophe modeling, or GIS analytics (insurance domain experience is a plus).
• Strong Python skills, including libraries like GeoPandas, xarray, rasterio, scikit-learn, NumPy, and pandas.
• Comfort working with large geospatial and time series datasets (e.g., flood zones, wind swaths, storm tracks).
• Understanding of machine learning concepts with ability to apply to insurance exposures, claims, or geospatial problems.
• Generalist mindset—comfortable navigating between data engineering, modeling, and product thinking.
• Strong communication skills, especially translating technical results into business insights.
• Use of modern tools including AI as accelerators, with independent validation and reproducibility.

Responsibilities

• Analyze and transform geospatial and time series data using Python-based pipelines.
• Build and support catastrophe model components, including hazard, vulnerability, and exposure layers.
• Conduct post-storm data analyses and collaborate with external stakeholders to deliver actionable insights.
• Create internal tools and reports that make complex datasets usable for business stakeholders.
• Collaborate with engineers and product teams to deliver insights that move the business forward.

Skills

Python

GeoPandas

xarray

rasterio

scikit-learn

NumPy

pandas

Geospatial analytics

Machine learning concepts

Communication

Tools

Docker

Kubernetes

Cassandra

S3

Job description

Are you a data scientist who thrives at the intersection of geospatial data, machine learning, and real-world impact? At Understory, we’re reimagining how weather and catastrophe risk are measured, understood, and priced—and we’re looking for a versatile Data Scientist to help us push the boundaries.

We work with rich sensor data, custom risk models, and a fast-moving engineering stack to power insurance products with precision and speed. If you’re comfortable jumping between modeling, mapping, and messy data, and you enjoy making your work actionable for underwriters and business partners, you’ll feel right at home here.

What You’ll Do:

• Analyze and transform geospatial and time series data using Python-based pipelines.
• Build and support catastrophe model components, including hazard, vulnerability, and exposure layers.
• Conduct post-storm data analyses and collaborate with external stakeholders to deliver actionable insights.
• Create internal tools and reports that make complex datasets usable for business stakeholders.
• Collaborate with engineers and product teams to deliver insights that move the business forward.

Your Day Might Include:

• Analyzing recent storm activity using data from our proprietary weather stations to support underwriting and quantify storm-related losses.
• Applying statistical and machine learning techniques to evaluate and calibrate the predictive performance of our catastrophe risk models.
• Conducting loss cost analyses to inform pricing, underwriting strategy, or reinsurance structuring.
• Writing and executing Python and SQL queries to analyze risk concentrations, portfolio statistics, and reinsurance projections.

What You Bring:

• 3–5 years of experience in data science, catastrophe modeling, or GIS analytics (insurance domain experience is a plus).
• Strong Python skills, including experience with libraries like GeoPandas, xarray, rasterio, scikit-learn, NumPy, and pandas.
• Comfort working with large geospatial and time series datasets (e.g., flood zones, wind swaths, storm tracks).
• An understanding of machine learning concepts (regression, classification, clustering, model validation, etc.) with the ability to apply them to insurance exposures, claims, or geospatial problems.
• A generalist mindset - you’re comfortable navigating between data engineering, modeling, and product thinking.
• Strong communication skills, especially when translating technical results into business insights.
• You use modern tools, including AI, as accelerators - not autopilots. You apply them thoughtfully to speed analysis, coding, and model development while independently validating results, owning your work, and maintaining a focus on statistical validity, reproducibility, and correctness.

Nice to Have:

• Familiarity with Docker, Kubernetes, or container-based workflows.
• Experience working with distributed/cloud-based data sources, especially Cassandra or S3.
• Experience running SQL queries and forming standard reports.
• Experience building tools or models for underwriters, risk analysts, or insurance decision-makers.

What We Offer:

We offer competitive compensation, full benefits, and the chance to do mission-driven work with tangible, real-world impact. You’ll be part of a small, collaborative team that’s helping redefine weather data and risk analytics in the insurance space.