Computer scientist @ University of London · Machine Learning Research Intern @ DeepKapha Labs
A context-free DDPM (Denoising Diffusion Probabilistic Model) training pipeline for wind-speed imagery using pre-generated NASA Global Precipitation Measurement (GPM) satellite data, combined with NASA Geostationary Operational Environmental Satellite (GOES) imagery RI data to identify intense convective bursts noise for fully deterministic and reproducible experiments.