At a prior gig we were using their FPGA instances and had a way of causing hard faults on the system that required reboots, with specific FPGA bitstreams. Their systems monitored this and when we reached out about it they scheduled at least 2-3 different calls with the engineers/PIs who were directly responsible so we could coordinate with them and help pin down the issue to get things working. We were literally renting by the hour; I don't even think we had an account manager, with variable costs; maybe a few hundred to 3k a month?
The funniest thing was that we got about a million in CAF bucks from them to go contract and move more stuff from AWS ( our monthly was low triple digits thousands ) and their AMs still messed everything up.
