Hash Maps & Data-Structure Design

If there is one coding round you can count on at Figma, this is it. The interviewer hands you a small piece of the product and asks you to implement its methods live on CoderPad, with tests that run. Build a document made of layers, each layer a bag of properties, with a way to change a property, read it back, and undo the last change. Under the surface it is a hash map and a stack, which is why so many candidates recognize the shape halfway through. The signal is not that you spot the structure. It is that you model the editor's rules correctly and keep the code clean enough to extend when the follow-up lands.

Figma's real document is a tree of objects, each with an id and a set of properties, held in memory and synced through a server that orders every edit. The interview version keeps the same spine and drops the networking. So the problems in this lesson all live in one family: a hash map from an id to a small structure, plus the operations an editor performs on it. Master the document-with-undo pattern first, then the same muscle carries you through an LRU cache of rendered tiles and an in-memory index over layer properties.