This year's fall enrollment Citadel As one of the top quantitative firms in North America, many students feel pressurized when they hear about Citadel! As one of the top quantitative companies in North America, many students would feel pressurized when they hear about Citadel, and OA must be difficult and tricky. But this experience can be said to be "too facetious", a total of two questions, a total of 19 minutes to do a steady finish, ideas and code are very smooth. I'm writing this down to share with you in the hope that it will give some reference to students who want to prepare for OA at Citadel or other big companies in the future.
Overall impression
- Number of questions: 2 questions
- Difficulty: medium-low
- Time: enough time limit (I finished in 19 minutes)
- Points: data structures (sliding windows), graph theory (topological sorting), heap optimization
If you've brushed up on the question pools on common OA platforms (e.g. HackerRank, LeetCode, CodeSignal) before, the ideas in these two questions aren't really new, and the main thing you'll be looking for is how quickly you can locate the correct solution and write a bug free implementation.
Problem 1: Detect Abnormal Transactions
Description:
You are given a list of stock trading records. Each record contains a timestamp and a trade volume. Determine whether there exists an " abnormal transaction," defined as a time window where the total volume exceeds a given threshold. abnormal transaction," defined as a time window where the total volume exceeds a given threshold.
Solution Idea:
- Use a sliding window to scan records in chronological order.
- Maintain the total volume inside the window by adding the new trade and removing the expired ones.
- If at any point the total volume exceeds the threshold, immediately flag it as abnormal.
Key Points:
- Optimize the sliding window update: don't recalculate the sum from scratch each time.
- Pay attention to edge cases, e.g., whether "equal to threshold" counts as abnormal or not.
This question is actually a sliding window template question, the logic is relatively intuitive, as long as the implementation is written concisely, there will not be too many pits.
Problem 2: Task Scheduling with Dependencies
Description:
You are given multiple tasks. Each task has an execution time and may depend on other tasks. A task can only be executed after all its dependencies are completed. Find the minimum total time required to complete all tasks, or output an optimal scheduling order.
Solution Idea:
- Build a directed graph based on task dependencies, compute in-degrees.
- Perform a topological sort to guarantee valid execution order.
- Use a min-heap to always select the task with the shortest execution time among the available ones.
- Accumulate the total time until all tasks are processed.
Key Points:
- Detect cycles: if the dependency graph contains a cycle, the tasks can't be completed.
- Dynamically update the available task set after finishing one.
- Handle edge cases such as single task, no dependencies, or purely linear dependencies.
This is a classic task scheduling problem with topological sorting + minimal heap. It's easy to fail if you don't get the details right, but if you're familiar with the template, it's relatively stable.
Summarize
This time Citadel 2026 Intern OA, two questions (abnormal transaction detection + task scheduling) overall friendly, 19 minutes can be smooth AC. overall feeling is to examine the basic algorithms and implementation skills, as long as the amount of questions enough, can be taken.
For those who want to attack Citadel or other major North American companies, accumulation of question types + practice of high frequency questions + simulation is definitely the key to victory. If you are well-prepared, the OA is just a warm-up session, and the real challenge lies in the VO and interviews later.
Specialized FacingThe whole process is escorted to the Offer.
Many students in the preparation of Citadel, Amazon, Google and these big companies, the most afraid of the most is not the question will not do, but the time is tight, the state of chaos, the details are easy to turn the car. We provide a full set of services from OA ghostwriting/Hackerrank to remote VO tips/interview assistance, all of which are operated by real experts with North American CS backgrounds, guaranteeing seamless and safe operation, and the results are far better than those of AI.
If you want to save more time, we also support professional face-to-face (including SDE, Quant, FAANG full coverage), with camera transfer and voice change technology, simulation test in advance to ensure tacit understanding, to help you escort all the way to the Offer.
Whether it's a single assist or a complete process (from OA to contract negotiation), we can customize it flexibly. Pay a small deposit in advance, and then pay the final amount after you get a satisfactory Offer, so we are truly result-oriented.
If you're also hitting the big North American factories, don't be a hard ass, and feel free to chime in if you need to!
