Postdoctoral Research Associate, Multimodal Machine Learning for Environmental Forecasting
I developed CNN-based and multimodal AI models to process and label satellite imagery and in situ sensor data for marine carbon removal and oceanographic event prediction research. This work involved engineering data fusion pipelines that integrated geospatial images and sensor data, requiring detailed annotation and segmentation of ocean events and features. The focus was on enabling robust AI/ML model training for environmental forecasting applications using large-scale geospatial datasets. • Labeled and segmented satellite and sensor data for use in deep learning climate models. • Worked with multi-terabyte environmental datasets using PyTorch and TensorFlow frameworks. • Created labeled datasets for algal bloom and sea ice thickness prediction. • Engaged with commercial stakeholders to ensure scientific rigor in data labeling for the Blue Economy sector.