Considering how requests are most likely not the part with most intense computations, we're not yet sure if it's Arkyo responsability to create and manage its threads.
With this in mind, a couple options could be considered for future implementations:
Use async in order to handle the requests. Arkyo may still become unresponsive if a function uses too much computation.
Use a thread/processing pool. Although it uses more memory, it may pose itself as a solution for Arkyo to not become irresponsive (except if the case of peak cpu usage).
Spawning a new thread/process and let it completely handle the request from then on.
Considering how requests are most likely not the part with most intense computations, we're not yet sure if it's Arkyo responsability to create and manage its threads.
With this in mind, a couple options could be considered for future implementations: