I had the problem:

TypeError: Object of type InferenceRequest is not JSON serializable

with code similar to this:

request = PandasCodec.encode_request(data)

endpoint = ""
response = requests.post(endpoint, json=request)

Since the requests library doesn't know how to serialise an InferenceRequest object, it is needed to convert it to a dict, which is the expected value of the json kwarg on requests.post. We can do this with request.dict().

It would be useful if we can document it better on this page:
https://mlserver.readthedocs.io/en/latest/user-guide/content-type.html

because I think the problem with input data / codecs is a big blocker for many people that want to try mlserver.