I have worked with a lot of Transformer based architectures ( BERT, ROBERTA, XLM, XLNET, T5, FLAUBERT, ALBERT and many more) over the past 18 months. Most of this has been possible using the Python based Huggingface Transformers open-source framework and it's derivatives. I have worked on downstream tasks like classification, named entity recognition, ranking, language generation , question answering etc using these Transformer based architectures.
As far as model deployment is concerned, any of the solutions like FastAPI, Django , Flask, AWS Sagemaker should do the job perfectly .
Thus for these reasons I think that I am the best man for your job.