I would investigate in two directions: The number of requests to the API depends on your pace and number of sub-requests.
How do you handle rate limiting on your part?
Hmmm. If you get a 429 you should wait and repeat. I see no reason that your API has to fail in that case.