Side note: local-global approximation seems to do well in graphics and visual tasks. For example in the field of alpha matting the state of the art for a while was KNN Matting which sampled locally and globally. Most methods since then have taken a similar approach.
most requests take only a fraction of a second to answer. to actually monitor threads on that frequency you'd need not only a significant portion for the monitoring (ever looked at the CPU usage with running top?) but also for the network communication to send that over to the lb, which needs to process the data for load balancing
Maybe it's just a cultural difference but there is something unsettling and cringey about seeing those employees act so enthusiastic.
Is it genuine? Maybe for some employees. But the camera setup makes me think there is some expectation placed upon the employees that they will be there to cheer. It's not natural.