I used curl to call transloadit.com's API so I know how the http body should look like. I wanted to use requests to import some images with transloadit. The key is to use requests PreparedRequest to manipulate the body.

gistfile1.txt

curl -F 'params="{"auth": {"key": "YOUR_KEY"}, "template_id"="YOUR_TEMPLATE_ID"}' \
-F "CUSTOM_FORM_FIELD=YOUR_CUSTOM_VALUE" \
-F "FANCY_THIS=FANCY_THAT" \
"http://api2.transloadit.com/assemblies"

gistfile2.txt

------------------------------3d01f393c9e7
Content-Disposition: form-data; name="params"

{"auth": {"key": "YOUR_KEY"}, "template_id": "YOUR_TEMPLATE_ID"}
------------------------------3d01f393c9e7
Content-Disposition: form-data; name="CUSTOM_FORM_FIELD"

YOUR_CUSTOM_VALUE
------------------------------3d01f393c9e7
Content-Disposition: form-data; name="FANCY_THIS"

FANCY_THAT
------------------------------3d01f393c9e7

transloadit.py

import json
from requests import Request, Session

TRANSLOADIT_URL = 'http://api2.transloadit.com/assemblies'

session = Session()

def import_image(auth_key, params):
    _params = {
        'auth': {
            'key': auth_key
        },
        'template_id': 'YOUR_TEMPLATE_ID'
    }

    params.update({ 'params': json.dumps(_params) } )

    prepared_request = Request('POST',TRANSLOADIT_URL, files=params).prepare()
    body = prepared_request.body
    for param in params.keys():
        to_replace = '; filename="%s"\r\nContent-Type: application/octet-stream' % param
        body = body.replace(to_replace, '')

    print body
    prepared_request.body = body
    response = session.send(prepared_request)
    print response.content


if __name__ == '__main__':
    params = {
        'CUSTOM_FORM_FIELD': 'YOUR_CUSTOM_VALUE',
        'FANCY_THIS': 'FANCY_THAT'
    }
    import_image('YOUR_KEY', params)