Bloomberg Coding Interview Questions

Return the longest stretch of a string that reads identically in both directions. This Bloomberg phone-screen staple expects expanding around every centre as the baseline, and the follow-up push points at the linear-time algorithm that reuses earlier radii. You will remember the even-length centres as well as the odd ones, handle ties by any valid choice, and say where the linear method earns its extra complexity. A subscription unlocks the full problem and the model solution.

Decide whether a loop of fuel stops can be completed and from which stop to begin, refuelling at each one before moving on. This Bloomberg question has an elegant single-pass answer: the trip is possible exactly when the total fuel covers the total cost, and the starting point is just after wherever the running balance hits its lowest point. You will prove that claim rather than assert it, and handle the case where no start works. Subscribers get the complete problem and the model solution.

Tackling complex pointer manipulation is the core of this popular Bloomberg interview question focused on hierarchical data structures. You are tasked with transforming a multi-layered linked list into a flattened, single-tier sequence while maintaining a precise depth-first traversal order and correctly updating all bidirectional connections. This exercise evaluates your dexterity with pointer logic, edge-case management, and structural invariants in custom collections. Gain access to the full problem statement and optimal verified solution by subscribing now.

Decide whether every distinct character of a string occurs equally often, then whether deleting exactly one character can make that true. This Bloomberg question escalates: the interviewer pushes for better than the naive try-every-removal approach, which means reasoning over the frequency-of-frequencies instead. You will build the counts once, enumerate the few shapes that admit a fix, and keep the extra space bounded by the alphabet. Subscribers get the complete problem and the model solution.

Decide whether a pile of sticks of fixed lengths can be laid out as a square using every stick exactly once and breaking none. This Bloomberg question is a partition search, and the whole game is pruning: sorting the sticks longest-first, rejecting totals that are not divisible by four, and refusing a side that already overshoots. You will place sticks into four buckets with backtracking and explain why the naive search times out without those cuts. Subscribers get the complete problem and the reference solution.

Learn which word most often follows each word in a training corpus, then answer lookups for any query word. This Bloomberg question is a bigram counting exercise with a precise contract: unseen words return nothing, and ties resolve alphabetically rather than arbitrarily. You will build nested counts in one pass over the sentences, decide whether to resolve the winner at training time or at query time, and keep the tie rule deterministic. The complete problem and the reference solution are available with a subscription.

Count the integers in each of several ranges whose decimal digits are all different from one another. This Bloomberg question invites a brute-force loop and then asks what happens when the ranges get wide, which is where digit-by-digit counting earns its keep. You will test the repeated-digit condition cheaply, answer many queries without redoing shared work, and handle ranges that cross a digit-length boundary. Subscribers get the full problem and the model solution.

Generate a self-referential integer sequence where each term reports how long ago the previous term last appeared, and return the term at a given position. This Bloomberg question is a simulation whose only subtlety is bookkeeping: the lookup table must record the previous term's position after the new term is computed, not before. You will keep last-seen positions in a map, handle the first appearance case, and reason about the memory the sequence needs. A subscription unlocks the full problem and the reference solution.

Design the classes behind a number-guessing game where the system reports whether each guess is high, low, or correct. This Bloomberg question is an object-design exercise, so the marks are in the separation: the secret, the comparison logic, the input loop, and the game state each belong somewhere specific. You will keep the randomness injectable so the game is testable, define when a round ends, and decide what an out-of-range guess does. The full problem and the reference design require a subscription.

Report every position where a chosen character appears in a string, except where the following position holds a particular marker. This Bloomberg warm-up is about reading the exclusion rule carefully and then implementing it in one pass with a safe look-ahead. You will avoid running past the end of the string, decide what happens at the final position, and state the complexity. The full problem and the reference solution require a subscription.

Send a group of people to two destinations at minimum total cost while filling each destination exactly half full. This Bloomberg question has a neat greedy answer driven by the cost difference per person, and then a follow-up that generalises to several destinations and breaks the greedy argument. You will sort by the difference, assign the clear cases first, and explain why the many-destination version needs a different technique. The full problem and the model solution require a subscription.

Build a stream that accepts numbered values in any order and releases them only once a gap-free run is available from the current position. This Bloomberg question is a pointer-and-buffer design, and the insight is that each value is emitted exactly once, so the total work stays linear despite the loop inside the insert. You will advance the pointer greedily after each insertion, return an empty result when the run is not yet complete, and keep the buffer simple. The full problem and the reference implementation require a subscription.

Count the non-empty subsets of an array whose elements add up to an even total, faster than enumerating every subset. This Bloomberg question turns on parity alone: only the count of odd values matters, because the even values can be included freely and the odd ones must be taken in pairs. You will derive the closed form from that observation, handle an array with no odd values, and avoid the exponential enumeration the statement rules out. A subscription unlocks the full problem and the derivation.

Remove as few brackets as possible so that a string becomes balanced, and return every distinct result achieving that minimum. This Bloomberg question is where breadth-first search beats backtracking: exploring one removal at a time level by level means the first level with any valid string holds all of them. You will deduplicate aggressively, validate in a single pass, and leave non-bracket characters alone. The complete problem and the reference solution are available with a subscription.

Pick out the words buildable from a small letter palette that also include one mandatory letter and reach a minimum length. This Bloomberg question is a filtering problem where a bitmask over the alphabet makes each word a single comparison instead of a character loop. You will encode each word as a mask, test containment and the mandatory letter together, and allow letters to repeat freely. The complete problem and the reference solution are available to subscribers.

Build a collection of addresses that supports insertion, deletion, and drawing one at random, all in constant time. This Bloomberg question has a specific intended answer: an array for random indexing plus a map from value to array position, with deletion swapping the last element into the vacated slot. You will keep the two structures in agreement, decide what a draw from an empty collection does, and reject duplicate insertions. A subscription unlocks the full problem and the reference implementation.

Assign every value to the interval between two adjacent boundaries and report how many landed in each. This Bloomberg question is binary search per value with a precise exclusion rule: anything below the first boundary or at or above the last is dropped rather than clamped. You will pick the search variant that matches the half-open intervals, handle values sitting exactly on a boundary, and keep the per-value cost logarithmic. A subscription unlocks the complete problem and the reference solution.