The best advice I ever got about interviews is to figure out ways to prove yourself wrong about what you think about the candidate. Although it's good advice for a candidate who seems too perfect, this advice really shines when someone is bombing.
Just take an unseen technical problem, solve it for the very first time in front of a colleague, then double your time as the expected time for candidates in an interview.
When you're under time-pressure, you actually spend more time thinking about how much time you have left, rather than thinking about how to solve the problem.
> Just take an unseen technical problem, solve it for the very first time in front of a colleague, then double your time as the expected time for candidates in an interview.
That gives very poor signal in my experience due to the time pressure. Your process is now selecting for people who are good at taking tests instead of selecting for good engineers.