There’s stupid simple way to test interview process - let your current engineers go through it. I bet best performing ones will fail mentioned algorithmic and abstract assignments.

Sometimes the best value to the team bring a person who doesn’t know algorithms but can e.g. foresee problems or ask “stupid” questions which lead others to think differently.