I don't know how it is done here but there are plenty of ways: VMs, sandboxes, etc... There is no problem running untrusted code as long as the I/O and interru

I don't know how it is done here but there are plenty of ways: VMs, sandboxes, etc... There is no problem running untrusted code as long as the I/O and interrupts are protected, something that is a standard feature of all the CPUs and OSes that the server may run on... Unless you find a bug to exploit.