Not able to understand where the extra time is getting added when a API is completed

Question

I am experiencing following weird behaviour -

From client(browser) a API call is made to server to fetch user details. The time required by this API to complete is around 650 ms. But in browser it is showing that the request got completed in 3 secs (this time varies from 1.5 sec to 5 secs).

I want to figure out where this extra time is getting added. I am not able to currently troubleshoot this problem. I know that there is some delay added due to network latency considering the client location and server location, but this delay is expected to be around 150 to 250 ms. But the difference in time that I am observing is too much.

Following are more details about the infrastructure -

Client (Browser) Location - Mumbai, India

Server Location - North Virginia (AWS Cloud)

On the backend we have micro services and these are deployed on AWS cluster and we are using Amazon Elastic Kubernetes Service. Our architecture looks like following -

ASW Load Balancer --> INGRESS --> NGINX --> API GATEWAY (Micro services)

I need to understand where this extra time is getting added, is it really network latency or there is some issue happening in the above architecture and if yes at which layer the delay might be getting added.