Dong Pham

dong_profile.JPG

PhD, research on large-scale remote sensing applications

Overview

Remote sensing scientist, PhD, specializing in continental-scale land-cover mapping and deep learning for Earth observation. I work at the EO Lab, University of Greifswald and focus on consistent land-cover mapping across large spatial and temporal domains. I bridge research and engineering by designing transferable models/datasets and deploying them on HPC and cloud platforms.

Highlights

  • BSRLC+ — Annual Europe-wide land-cover dataset (2000–2022) with cropland and peatland detail; introduced Temporal Encoding for irregular EO time series. Learn more · Method
  • BSRLC-U — Tri-annual 10-m built-up type maps over two decades using Landsat + Sentinel-2; combined super-resolution and center-patch classification. Learn more

Expertise

  • 6+ years building continent-scale land-cover products on FORCE datacubes and cloud platforms. Report
  • Deep learning for EO time series; cross-sensor/year transfer; fractional mapping; super-resolution.
  • Production pipelines on Slurm/HPC and cloud (e.g., CODE-DE).
  • Domains: Urban, Forests, Croplands, Wetland vegetation.

Technical skills

Programming & Data: Python (pandas, rasterio, PyTorch, matplotlib); packaging, testing, CLIs · JS/TS (React, Leaflet) for web viewers · Bash · Git/GitHub (CI/CD)
ML & Time Series: PyTorch (CNNs/Transformers, 1D/2D CNNs, LSTM); TensorFlow/Keras for prototyping
Remote Sensing & GIS: GDAL/rasterio; FORCE datacube report; CDSE OData API; author of CDSE-S2 Downloader code
Data Engineering & Platforms: Containers (Docker/Singularity); Slurm on HPC; cloud deployments on CODE-DE and related platforms

Publications

Full list: publications.