I’ve sat on both sides of that table for two decades. I’ve asked gotcha questions that were trivia, not skill. I’ve been on the receiving end of the same. I have also run interviews I knew were worse than they should have been.
If you are hiring a senior engineer, you need two things: a sharp picture of what the role actually has to do, and a way to see whether this person can do it. Charm is not a proxy for either.
Some people interview beautifully. They tell a clean story. They sound like they belong on a stage. That is a skill. It overlaps only loosely with the person who will still be useful six months after the offer.
If you cannot say what "great" means for this seat, the charismatic candidate will define it for you.
Have them teach you something
If you need a mentor, do not ask "tell me about a time you mentored someone." That prompt has a canned answer.
Give them five minutes. Let them pick something they know — a design pattern, a debugging trick, a cloud quirk — and teach it to you. You learn how they break complexity down. You see patience or the lack of it. You get a read on depth without pretending you share their stack.
Walk a real problem, then break it
If you need someone who can untangle a hard system, have them walk a project they owned. What was the problem? What did they decide? Then change the facts. Mid-project, the requirement flips, or a resource disappears. Watch them think. A rehearsed war story falls apart. A real problem-solver starts drawing.
Talk about values like an adult
If culture matters, say what yours actually is. Then ask which value they would lean on in a senior seat, and why. "Yes, I agree with those" is not a signal. A specific story is.
What drives seniors away
A process they cannot see. Surprise rounds, vague timelines, a role nobody can describe. Senior people will not gamble on a job you cannot explain.
HackerRank and weekend take-homes. For juniors the debate continues. For someone with fifteen years, a generic algorithm puzzle says you do not value their time, or that you think the job is typing. AI can already do the boilerplate. Ask about architecture and judgment.
Starting with a timed whiteboard puzzle. If the conversation is good, you will end up at the board anyway — to sketch a system, not to perform.
Interviews are already stressful. If you add trick questions and an adversarial tone, you are advertising the job. The people you want will leave.
You are not lowering the bar. You are stopping the process from selecting for performance.