How do we spot who'll ramp up versus flame out?

You think the problem is your judgment. It isn't. The problem is you have no written standard for what 'ramping well' looks like, so every hire is a fresh guess, and every guess feels like judgment because you have nothing else to call it. A resume tells you what someone did elsewhere, for someone else, under conditions you don't share. An interview tells you how well they talk about work. Neither tells you how they will act under your conditions, in your first ninety days. You already knew this. You hired anyway, hoping instinct would cover the gap. It won't, and it hasn't. The fix is not a better interview question. It is deciding, before anyone starts, exactly what actions prove someone is learning and adapting — then watching for those actions instead of a feeling. Do this with your team, not alone. Four eyes catch what one pair misses, and a shared standard stops arguments from becoming guesswork later.