5 months ago
Remote, United States or Toronto, CanadaStaff+
Responsibilities
- Architect and build machine learning models that map and predict vegetation and fuel conditions across diverse geographies.
- Design and maintain data and feature pipelines for large-scale geospatial and temporal data.
- Partner with wildfire science and product teams to define modeling objectives and real-world evaluation metrics.
- Build reproducible experimentation frameworks and model evaluation workflows.
- Scale models from research to production with emphasis on performance, reliability, and explainability.
- Lead the evolution of ML systems, tooling, processes, and modeling standards.
- Collaborate with MLOps peers on production training, inference, and monitoring.
- Drive architectural decisions and mentor other engineers.
Requirements
- 10+ years of experience designing and building production-grade machine learning pipelines and systems.
- Experience across machine learning, geospatial data, and environmental science.
- Strong background in deep learning, computer vision, or remote sensing.
- Experience designing end-to-end ML systems from data ingestion and preprocessing through deployment and monitoring.
- Hands-on experience with PyTorch, TensorFlow, XGBoost, LightGBM, Dask, Spark, or GeoPandas.
- Familiarity with GCP and Vertex AI or similar cloud-based ML platforms.
- Strong communication and cross-functional collaboration skills.
- Ability to lead architectural discussions and mentor engineers.
- Preferred experience in wildfire science, forestry, or remote sensing.
- Preferred experience integrating physics-based models with ML, active learning, or uncertainty quantification.
- Preferred experience with model interpretability, data provenance, and deep learning models for weather or climate data.
- Experience in remote-first or globally distributed teams is a plus.
Benefits
- Remote work from one of the listed countries: United States, the Netherlands, United Kingdom, Ireland, Estonia, Portugal, France, Sweden, Switzerland, Denmark, or Canada.
- Eastern North America time zone requirement.
- Home office stipend, coworking budget, and ongoing education budget.
- Flexible, autonomous, and collaborative work environment.
- Annual in-person team gathering with occasional in-person collaboration.
- Competitive, location-specific benefits.
Tech Stack
Categories
Data EngineeringML Engineering
About Overstory
Overstory builds AI-powered vegetation management and grid resilience software for electric utilities, combining satellite imagery and machine learning to identify and prioritize vegetation risks. It sells its platform and analytics on a subscription and services basis to utilities and grid operators to reduce outages and wildfire risk. Founded in 2018 and headquartered in Somerville, Massachusetts, the company is privately held.