Sean Ridge Email and Phone Number

- Responsible for evaluating computer vision model performance, and implementing performance enhancements by improving training datasets or changing modeling approach- Created the Wildfire Vulnerability Score model and its training dataset consisting of geospatial data from damage surveys and computer vision extracted features from aerial imagery. - Increased speed of computer vision model evaluation by 10x by implementing parallelism- Accelerated annotation process and computer vision dataset creation by creating pipeline for transforming raster labels/predictions into PostGIS geometries

- Built “Weather or Not”, a webapp (FastAPI) issuing improved forecasts of weather-driven NYC gridlock- Created ETL pipeline for messy traffic speed data from a citywide sensor network- Trained an LSTM in TensorFlow to predict traffic delays using traffic and weather data- Deployed the model with Docker (github.com/sridge/weatherornot)

- Improved our understanding of climate projections of ocean CO2 absorption by building and analyzing ocean models of varying complexity- Studied impacts of climate variability (NAO) on ocean CO2 absorption- Used xarray and dask to process terabytes of climate simulation output for research projects- Helped peers run simulations with the Weather Research and Forecasting (WRF) model and analyze NCEP reanalysis data