I am currently working with AWS. I use an EC2 instance with Opensearch installed on it.
My client uses its own AWS account, with Amazon Opensearch Service (managed Opensearch).

We both use the same lambda to compute scores, and the same input data. The lambda calls a library named “Elasticsearch”.

However, we get different scores for the same queries, and this has an impact on the sorting of answers.

Can someone explain to me why we have this phenomenon ? Is it because I use an EC2 and the client Managed Opensearch ? Is it because we do not work with the same AWS accounts ?