I'm the CTO at Zamzar and we have an API to do just this available at https://developers.zamzar.com/
We have a Test account you can use for free to try out the service, and code samples for Python in our docs which which would enable you to convert a PDF file to DOCX quite simply:
import requests
from requests.auth import HTTPBasicAuth
api_key = 'YOUR_API_KEY'
endpoint = "https://sandbox.zamzar.com/v1/jobs"
source_file = "/tmp/my.pdf"
target_format = "docx"
file_content = {'source_file': open(source_file, 'rb')}
data_content = {'target_format': target_format}
res = requests.post(endpoint, data=data_content, files=file_content, auth=HTTPBasicAuth(api_key, ''))
print res.json()
You can then poll the job to see when it has finished before downloading your converted file.