A process-based pmap with sensible defaults, ready to go

pmap.py

from multiprocessing import Pool

def pmap(f, collection, size=10):
    """
    Applies `f` in parallel over `collection`.
    Pool `size` has a sensible default of 10.
    """
    with Pool(size) as p:
        return p.map(f, collection)

And the Threaded version, useful on AWS Lamda, where processes are restricted:

from concurrent.futures import ThreadPoolExecutor

def tmap(f, collection):
    """
    Applies `f` in parallel threads over `collection`.
    Pool `size` has a sensible default of 10.
    """
    with ThreadPoolExecutor(max_workers=10) as executor:
        return list(executor.map(f, collection))