3 ms·

Running risk-management models on rando client-side boxes seems like an amazing way to test your algorithm's ability to tolerate Byzantine failures.