AI Wildfire Risk Prediction Engineer

Paleon Data

Remote · $130k – $160k / year · Freelance · Posted 4 months ago

This role is no longer accepting applications

Don't miss the next role like this

Create a free profile with your skills. We match you to new AI roles as they open, and companies hiring can contact you directly.

Create free profile, 2 minutes

Similar open roles you can apply to now:

See all open AI jobs

This role is no longer accepting applications

Don't miss the next role like this

Create a free profile with your skills. We match you to new AI roles as they open, and companies hiring can contact you directly.

Create free profile, 2 minutes

Similar open roles you can apply to now:

See all open AI jobs

About the role

The models you'll ship will influence evacuation decisions for 2.3 million people in high-fire-risk zones across the Western US. County emergency management offices, utility companies making grid de-energization decisions, and insurance carriers updating coverage policies all rely on our risk assessments. Getting them wrong has real human consequences — in both directions. We need a freelance ML engineer to rebuild the core risk prediction model. The current system uses fuel moisture, wind speed, and topography — standard variables that every platform uses. We want to incorporate satellite-derived live fuel moisture indices, historical fire perimeter data, prescribed burn records, and fire weather indices into a spatiotemporal model that generates 72-hour risk forecasts at 1km grid resolution. This is a 9-month engagement, 30-40 hours per week. You'll have access to petabytes of satellite and climate data on GCP. The science team provides domain guidance; you own the model architecture and training pipeline.

Fully remote.

Responsibilities

Design and train a spatiotemporal wildfire risk prediction model at 1km resolution
Integrate satellite-derived fuel moisture indices and historical fire perimeter data
Build data pipelines for multi-source climate and remote sensing inputs on GCP
Develop 72-hour ensemble forecast generation with calibrated uncertainty estimates
Evaluate model performance against historical fire events and expert risk assessments
Deliver weekly progress reports and model performance dashboards

Requirements

6+ years in geospatial ML, climate modeling, or environmental data science
Experience with spatiotemporal modeling and remote sensing data
Strong Python, PyTorch, and distributed computing skills on GCP or AWS
Familiarity with fire weather indices and fuel moisture modeling is a strong plus
Ability to work independently with a defined research agenda