Citadel Coding Interview Questions
25 Citadel coding interview problems with full optimal solutions — 8 easy, 12 medium, 5 hard. Every problem ships with multiple approaches (brute-force first, then the optimal), complexity tables for each, company-specific tips on what an Citadel interviewer values, and a FAQ section.
Showing 12 problems of 25
- #3mediumfrequently asked
3. Longest Substring Without Repeating Characters
Sliding window on a character stream is a core primitive in financial data processing — think deduplicating a stream of order IDs, finding the longest non-redundant price sequence, or scanning for unique symbol windows in tick data. Citadel uses this problem to verify you can implement an O(n) shrinkable window correctly.
- #15mediumfrequently asked
15. 3Sum
3Sum tests whether you can compose a known primitive (Two Sum) into a harder problem and handle duplicate suppression correctly. Citadel often uses it as the natural escalation after Two Sum, specifically probing your O(n²) two-pointer approach and whether you can articulate why sorting is the enabling step.
- #49mediumfrequently asked
49. Group Anagrams
Group Anagrams is a hashing canonicalization problem: find a transformation that maps equivalent items to the same key. At Citadel, this pattern appears in symbol normalization (mapping ticker aliases to a canonical identifier) and in deduplication pipelines. The interview tests whether you choose the right canonical key — sorted string vs. frequency count.
- #56mediumfrequently asked
56. Merge Intervals
Interval merging is a direct analog of consolidating overlapping trading windows, computing continuous market hours, or merging order-book time slots. Citadel uses this problem to test sort-first reasoning and the ability to handle all overlap cases correctly — especially the 'contained within' edge case.
- #139mediumfrequently asked
139. Word Break
Word Break is a DP problem about partitioning — given a string and a dictionary, can the string be segmented into valid words? At Citadel, analogous problems appear in pattern recognition on order strings and in tokenizing structured text streams. The interview tests DP state definition and whether you avoid re-checking overlapping subproblems.
- #146mediumvery frequently asked
146. LRU Cache
LRU Cache is directly relevant to Citadel's systems work: caching market data, reference data (security master), and computed risk metrics all require eviction policies. Citadel interviewers use this to verify you can compose a hash map and doubly-linked list into O(1) get and put — and that you understand why each structure is necessary.
- #200mediumfrequently asked
200. Number of Islands
Number of Islands is a graph traversal problem disguised as a 2D grid. At Citadel, graph connectivity appears in portfolio risk analysis (connected components of correlated positions) and in network topology modeling. The interview probes BFS vs DFS choice and whether you can modify the grid in-place to avoid a visited set.
- #207mediumfrequently asked
207. Course Schedule
Cycle detection in a directed graph is a fundamental primitive in dependency resolution — trade execution sequencing, pipeline DAG validation, and margin calculation dependency graphs all require verifying no circular dependencies exist. Citadel expects both DFS (with coloring) and topological-sort (Kahn's) approaches.
- #238mediumfrequently asked
238. Product of Array Except Self
This problem prohibits division and demands O(1) extra space (excluding output) — a tight constraint that forces a prefix/suffix product pattern. At Citadel, this mental model maps to computing rolling window statistics on time-series data while excluding the current observation. The constraint bar is a deliberate filter.
- #295mediumvery frequently asked
295. Find Median from Data Stream
Computing a running median over a live data stream is a real operational metric at Citadel — median latency of order acknowledgements, median fill rate over a rolling window, median bid-ask spread. The two-heap technique is the standard interview answer, but Citadel pushes hard on the rebalancing logic and on what happens at O(log n) vs O(n).
- #322mediumfrequently asked
322. Coin Change
Coin Change is the canonical unbounded knapsack problem and a gateway to understanding DP on finite targets. At Citadel, structurally identical problems appear in portfolio replication (can you replicate a target payout using a minimum number of instruments?). The interview tests whether you define the DP state precisely and initialize correctly.
- #347mediumfrequently asked
347. Top K Frequent Elements
Top-K selection shows up constantly in quantitative finance: top-K movers by volume, top-K securities by realized volatility, top-K events by arrival rate. Citadel expects you to know all three approaches — sort, min-heap, and bucket sort — and choose the right one based on the relationship between n and k.