The round: a frequently reported Uber object-oriented design question (eight reports here), in phone screens, onsite coding rounds and one AI-assisted coding round. The reported core: a parking lot with park, unpark and checkCar, where motorcycle spots take only motorcycles and regular spots take both motorcycles and cars. Another report names the operations check-in and check-out; another calls it a parking building. In one screening round the candidate first designed the high-level structure with black-box components, then was asked for fully runnable code, with the focus on data structure choice, code quality and performance. One candidate who had not prepared it did not finish the code. Two graded parts here: one lot, then a multi-level building that stays fast at scale.

Part 1 — park, unpark, check_car, with spot types

class ParkingLot:
    def __init__(self, spots: list[str]): ...                     # spots[i] is "motorcycle" or "car"
    def park(self, plate: str, vehicle: str) -> int | None: ...   # vehicle is "motorcycle" or "car"
    def unpark(self, plate: str) -> int | None: ...
    def check_car(self, spot: int) -> str | None: ...

• A "motorcycle" spot takes only motorcycles. A "car" spot takes cars and motorcycles.
• park returns the spot index it used. A car takes the lowest-numbered free car spot. A motorcycle takes the lowest-numbered free motorcycle spot, and only if there is none, the lowest-numbered free car spot. If no allowed spot is free, return None.
• unpark frees the vehicle's spot and returns its index, or None if that plate is not parked.
• check_car returns the plate in a spot, or None if the spot is empty.
• Raise ValueError for an unknown vehicle type, a plate that is already parked, or a spot index that does not exist.

Part 2 — a multi-level parking building, fast at scale

class ParkingBuilding:
    def __init__(self, levels: list[list[str]]): ...    # levels[l][i] is the type of spot i on level l
    def park(self, plate: str, vehicle: str) -> tuple[int, int] | None: ...
    def unpark(self, plate: str) -> tuple[int, int] | None: ...
    def check_car(self, level: int, spot: int) -> str | None: ...

• Same rules as Part 1. Spots are identified by (level, index), and "lowest" means lowest level first, then lowest index. A motorcycle still takes any free motorcycle spot in the building before a car spot.
• park and unpark must stay fast: the tests fill 200,000 spots and then run 100,000 departures and arrivals. Scanning for a free spot on every park is too slow.

Notes: the reports give the operations and the spot rule, not an interface. The class and method names, string vehicle types, spot numbering, "lowest free spot first", a motorcycle preferring a motorcycle spot, None for a full lot or an unknown plate, and the ValueError cases are ours. The reference also accepts check_in and check_out as names for park and unpark.