Just finished Citadel Summer 2026 SWE Internship Online Assessment, both programming questions were successfully passed! The overall difficulty is medium to high, and the time limit is about 70 minutes to complete. The questions test the flexible use of basic data structures, boundary processing, and time complexity optimization. The platform is HackerRank, a classic two-question format. Let me share two questions I encountered, as well as my personal thoughts and experience summary. I hope it will be helpful to students who will brush Citadel OA 2026 in the future!
Problem 1: Count Stable Subsegments
In an organization, there are n servers with capacities given in an array capacity. A contiguous subsegment [l, r] (with r – l + 1 >= 3) is considered stable if capacity[l] == capacity[r] and this value equals the sum of capacities from l+1 to r-1.
Return the total number of such stable subsegments.
Idea sharing: The essence of this question is to examine the conditional judgment of a subarray. Direct brute force enumeration of all possible l and r (ensuring length >=3) is feasible, but requires optimization of intervals and calculations.
- First calculate the prefix sum prefix[], so sum(l+1 to r-1) = prefix[r] – prefix[l+1].
- A double loop goes through all l and r (r >= l+2), checking that capacity[l] == capacity[r] and is equal to the intermediate sum.
- Note the bounds: when the length is exactly 3, the intermediate sum is one element; longer lengths are calculated normally. The time complexity is O(n²), n is generally within 10^5 and it is enough to pass. The key point is to handle the situation between the index of the prefix sum and the space (although there is no space for length >= 3).
This question has many boundaries, and it is easy to get stuck in sum calculation or indexing. It is recommended to test a few more examples such as [3,3,3] and [9,3,3,3,9].
Problem 2: Collect Goodness of Strictly Increasing Subsequences
Given an array of integers, consider all possible strictly increasing subsequences (including single elements). For each such subsequence, compute its “goodness” value through cumulative bitwise OR operations along the subsequence.
Collect all unique goodness values from all strictly increasing subsequences, and return them in sorted order.
Idea sharing: This question examines the combination of subsequence and bit operations that requires dynamic maintenance of possible goodness values.
- Use a set to maintain all possible current goodness values (0 can be added initially, if empty subsequences count).
- To handle strict increments, you need to keep track of the last element corresponding to each goodness (you can use a map or a set of pairs).
- When traversing the array, for the current nums[i], traverse all states in the current set. If nums[i] > last_val of the previous element, calculate new_goodness = old_goodness | nums[i] and add to the temporary set.
- After traversing the current element, merge the temporary set into the main set, and update last_val to nums[i].
- Finally, the set is converted into a sorted list and returned.
Key optimization: avoid explosive growth and use set to remove duplicates; if last_val is not tracked, strict increment conditions will be missed. The time complexity depends on the number of states, but in actual cases it is controlled within a reasonable range.
In summary, the question style of this round of Citadel OA 2026 is quite typical: one subarray + prefix sum counting, one subsequence + DP-like state maintenance + bit operations. The difficulty is close to LeetCode Medium-Hard. It is recommended to focus on questions related to arrays/prefix sums, DP, bit operations, and subsequences.
Why do 90% of people get hung up on Hidden Cases?
Citadel's OA system is tricky.
- Timeout trap: Q1 If the prefix sum is not used, or Python is written too slowly, it will fail when dealing with large amounts of data.
- Boundary conditions: In Q2, what if the array is empty? What if they are all negative numbers? What if they are all repeated numbers?
- Code Style： Even if you can pass, if the code is written like an "intern" instead of an "engineer", it will still be rejected during the Hiring Manager Review stage.
We provide not just answers, but tickets to Offers.
- Real-time Assistance: Our Senior Engineer will guide you in coding in real time through screen sharing. Instead of sending you code to copy and paste, we let you watch us write, or we teach you how to "Think Aloud" through voice, making the interview process seamless.
- Original Code: Every line of code is written live by a real person, and it definitely passes CodeSignal’s Plagiarism Check.
- Safety First: We not only guarantee results, but also safety. The whole process simulates the real rhythm of answering questions, rejects writing in seconds, and rejects mechanized operations.
The recruitment window for Citadel 2026 is extremely short. Are you still anxious about completing LeetCode questions? The current employment environment, opportunities > strength, speed > perfect.
Contact ProgramHelp now Send the password [Citadel OA 2026].
