Trees, Graphs & Search

Trees and binary search are the quiet middle of the PayPal loop. The object-modeling round is the differentiator, but these show up often enough that a shaky level-order walk or an off-by-one in a rotated search will end a screen that was going fine. The bar sits at a solid medium. Nothing here is exotic. What the loop checks is whether you can write a clean traversal, keep the boundary cases straight, and say why your indices are where they are.

Most of this cluster runs off one skeleton: a depth-first or breadth-first walk that remembers where it has been. A tree lets you drop the visited set, because it has no cycles and every node has one parent. A graph or a grid puts it back, since two paths can reach the same cell. Binary search is the odd one out, a different discipline, and it gets its own section below. It fails at PayPal in the places the hidden tests are built to catch: the empty input, the single element, the index one past the end.

The PayPal angle

PayPal grades the boundaries as hard as the algorithm, because a payments company treats a wrong edge case the way it treats a wrong amount. The HackerRank online assessment is known for hidden tests that target the empty tree, the single node, the off-by-one on a search window. Clear those out loud before you code and you are already ahead of most candidates; two more habits fit this round. Java is the default here, so write it in a modern style. If your screen is a Karat one, expect the rapid-fire opener to ask the Big-O of a tree or a hash lookup, then comes a single problem that layers on extensions. Narrate throughout. Going quiet is scored against you directly.