- parallelize data downloading (Ty)
- automate creation of sub-area geojson based on min/max values from UNOSAT data
- decide on MPI vs mapreduce vs whatever else
- create a diagram/map of how code relates to one another
- identify steps to take when pre-processing images
- train a trial NN for classification
- create a workflow to check if a point from UNOSAT data is in a given geojson geometry with geopandas (look at spatial join functions)