Machine Learning Engineer

ABOUT FLOODBASE 

Floodbase is changing the way businesses and communities adapt to increasing risk of climate change impacts from flooding. Last year, we launched a solution enabling re/insurers and public sector organizations to profitably design, underwrite, and monitor parametric flood insurance products, extending coverage to historically uninsurable locations and risks. Built on nearly a decade of industry-leading and peer-reviewed science, our proprietary solution continuously monitors flooding globally. Floodbase is backed by investors like Lowercarbon Capital, Collaborative Fund, and Floating Point, and trusted by NASA, The UN, Google, FEMA, and more.

About the Role

We are looking for an experienced Machine Learning Engineer to build algorithms to synthesize data from satellites, hydrologic models, and ground sensors into actionable insights. We produce flood data that informs emergency response and enables parametric flood insurance for governments, farmers, and businesses. You will take ownership of large projects, and your work will shape the full pipeline from training data selection to model training, validation, and output transformation and analysis. You will work with a team of scientists and engineers with expertise in remote sensing, hydrology, index insurance, climate, social vulnerability, and machine learning to build the next generation of tools to ensure financial protection from floods. We produce cutting-edge AI climate science, and you will play a large role in leading the edge of this field, with opportunities to publish in leading journals and present at top tech and science conferences. The role is remote, with core hours in EST or hybrid based in Brooklyn, NY.

 

Here’s what you’ll do

• Use your analytical expertise to derive robust insights from large structured and unstructured datasets of observations, hydrologic model outputs, and auxiliary records (stream gauges, water occurrence, …) that can be incorporated into the Floodbase parametric insurance product
• Example project: Automate the manual validation process of our historical flood data for individual areas of interest (AOIs) by finding robust rules leveraging multiple validation data sources.
• Estimate risk of flooding and its associated uncertainty bounds informed by satellite observations for parametric insurance product design in collaboration with insurance market experts and product managers
• Example project: Find ways to add uncertainty estimates to our flood parametric product by sampling additional data points spatially around the AOIs or with statistical resampling methods like bootstrapping.
• Utilize your proficiency in Machine Learning to advance the state of the art in flood detection from large datasets of observational data and models that can specifically inform the flood response needs of Floodbase customers
• Example project: Train a global flood segmentation model on Sentinel-2 data that improves on challenging urban scenes and inside mountain and cloud shadows.
• Be a technical leader in the team and advocate for customer-centric product development based on time-bound scientific hypotheses-oriented experiments. 
• Manage, mentor, and recruit exceptional data scientists from a variety of backgrounds
• Visualize and explain work through presentations and notebooks
• Work closely with our team of machine learning engineers, research scientists, and software developers to translate rigorous science and research into robust products

Who you are

• An intellectually curious and driven individual who is interested in working on innovative, impactful problems to address the climate crisis
• Bachelor’s in a quantitative discipline with 4+ years of industry experience or Master’s or PhD with 1-2+ years of industry experience.
• Have demonstrated experience working in interdisciplinary teams solving complex problems that rarely have a clean, correct solution
• High coding proficiency in Python: can independently develop code and make and review contributions to a shared repository. 
• Familiar with open-source geospatial ecosystems (rasterio/gdal, shapely, geopandas, xarray/rioxarray, dask)
• Mastery of the fundamentals of scientific data analysis, statistics, and machine learning 
• Have a proven track record (open source projects, Kaggle, publications, etc) of developing new models for innovative applications
• Keep up with state-of-the-art modeling techniques, like foundation models, and identify practical approaches for scalable solutions to environmental problems.
• Prior experience with satellite imagery and/or geospatial modeling is a plus