IRT Ascend: Climb to Mastery

The same 2-parameter-logistic scoring model behind real certification exams, rendered as a landscape: the height at each point along the ability axis is the real log-likelihood of the candidate's answers so far, and a marker performs genuine gradient ascent — climbing toward the peak, the maximum-likelihood ability estimate — as each new item is answered with a real Bernoulli draw from the item's 2PL probability curve. Watch the terrain sharpen and the marker converge on the candidate's true hidden ability, and see whether that estimate clears the certification pass threshold.