See the ArcGIS Desktop help topic on Listing data for how to access feature classes (shapefiles) inside workspaces (subfolders).
Generally, you will need to set the main env.workspace, loop through
folders = arcpy.ListWorkspaces('*', 'Folder'), reset the env.workspace to the 'folders' list item, and use another loop
fcs = arcpy.ListFeatureClasses(wildcard), where 'wildcard' corresponds to your specific shapefiles (which can also be a list that is looped through). Tends to be slow...
If you would like to use a faster python module for listing the subfolders try os.listdir. And a somewhat similar post to yours can be seen here using os.walk.
My favorite is glob with the relevant wildcard (e.g. '*.shp') - a subfolder/sudirectory example is shown in this PyMOTW - glob. Below is code specific to your shapefiles:
main_dir = r'C:\Testing\location'
# Loop through all files by wildcard
for wildcard in ['POIS.shp', 'STREETS.shp']:
# Specify the subdirectory variable and specific shapefile name to match
sub_dir = main_dir + '/*/' + wildcard
# Loop through files in subdirectory
for in_fc in glob.glob(sub_dir):
# Specify output variable
out_fc = in_fc.rstrip('.shp') + '_Surrounding.shp'
# Buffer - insert your own parameters here
arcpy.Buffer_analysis(in_fc, out_fc, "100 Meters", "FULL", "ROUND", "ALL")
You could use an attribute field for the distance value or specify this using a dictionary (instead of the wildcard list in the example) in case you need different buffer distances for each shapefile type.