OOD: Random Bingo Card Generator

Onsite OOD coding round. Generate a random 3x9 Bingo board where each row has exactly 5 numbers and 4 blanks, each column is constrained to a decade range (1-10, 11-20, ...), and cells are filled probabilistically while keeping the row/column constraints feasible. Requirements Generate a random Bingo board of fixed size 3 rows x 9 columns. Each row contains exactly 5 numbers and 4 blanks (zeros). Each column is constrained to a value range, left to right: column 0 holds 1-10, column 1 holds 11-20, column 2 holds 21-30, and so on. Fill the board cell by cell. When filling a cell, respect how many numbers / blanks the row already has, and choose probabilistically whether to place a number or a blank so that the per-row (5 numbers, 4 blanks) and per-column constraints all remain satisfiable. Notes The hard part is keeping the global constraints feasible while filling greedily: track the remaining numbers needed per row and remaining capacity, and bias each random choice so you never reach an infeasible state (e.g. a row that still needs 3 numbers but has only 2 unfilled cells left). Numbers within a column usually appear in increasing order top-to-bottom in standard Bingo — clarify whether ordering within a column is required. This belongs to a small family of Uber random-board OOD prompts (Bingo board, minesweeper board) where the signal is constraint-aware random generation, not the data structure itself. Preparation Implement the generator and verify by sampling many boards that every row has exactly 5 numbers and every column stays within its decade range. Practice the "remaining-need vs remaining-capacity" feasibility check so random choices never violate the row / column quotas.