Koushik Ghosh

I build Python geospatial pipelines using Google Earth Engine and Microsoft Planetary Computer to produce monthly crop-sowing reports for India’s PMFBY crop insurance scheme. I train CNN classifiers and LSTM models on Sentinel-2 vegetation-index time series to automate crop-sown detection and sowing-stage prediction.

Previously at Vassar Labs IT Solutions, I delivered agricultural, hydrological, and terrain analysis workflows spanning sown-area estimation, crop-pattern classification, watershed delineation, catchment mapping, and drainage extraction. I worked with developers and data scientists on AI/ML models for land-use classification, crop identification, and change detection, while supporting client training and technical documentation.

I’ve also produced crop assessment and vegetation-dynamics maps at NRSC-ISRO and co-authored three peer-reviewed remote-sensing publications, including work published in Nature Scientific Data and Nature Scientific Reports. I’m focused on applying Earth observation, satellite time series, and GeoAI to agriculture and climate risk