Ground your plan in what you can demonstrate today.
A resume tells us what you have touched. A diagnostic tells us what you can explain, apply and reason through today.
A conversation, not a questionnaire.
Follow-ups probe depth and confidence. A shallow answer and a detailed one lead to different next questions.
Question
How do you decide whether a metric change in an experiment is real?
Answer
I would check the p-value and ship if it is under 0.05.
Follow-up from SPHERO
What if you looked at twenty metrics and only this one crossed the line? How would that change your call?
Evidence update: Experiment design: still needs evidence. Sphero probes multiple testing and practical significance.
Sample exchange. The follow-up depends on the answer, not a fixed list.
Four kinds of question.
- Concept
Can you explain it plainly and correctly?
- Applied reasoning
Can you use it on a realistic problem?
- Trade-off
Do you know what you give up with each choice?
- Role-round simulation
Only where your target process includes that round.
What you leave with.
- A grounded capability map
- Confirmed gaps
- Uncertain areas
- Your first next-best action
Rules of the room.
- You can pause and return. Voice or video is optional.
- Only role-relevant demonstrated capability is evaluated. No grade on your intelligence.
- No emotion or face analysis, and no unsupported psychological inference.
- You can inspect the evidence behind any result and challenge it.
- Rubric dimensions: Correctness, Reasoning, Trade-offs, Structure, Communication.
- What is captured and how long it is kept is stated before you begin.