← Back to blog Goldman CoderPad OA: Boxes, Subarray, Math, HireVue
OA Goldman Sachs

Goldman CoderPad OA: Boxes, Subarray, Math, HireVue

2026-06-12

Goldman Sachs interviews have always had a reputation for being "hard"—the Behavioral part digs into details, and the Technical is graded strictly, leaving almost no room for luck. This piece is organized from a recent GS OA + VO debrief by an oavoservice student, walking through the OA's 2 coding questions, 9 math multiple-choice, and the HireVue prompts, plus how the OA assist / OA live support path plugs in.

1. Overall OA structure

This round of the GS OA is a bit of a "mashup":

Module Content Time
Coding 2 CoderPad-style questions Untimed, but think out loud while coding
Math MCQ 9 questions, probability / linear algebra / calculus mix ~40 minutes
HireVue 6 recorded behavioral prompts 45s prep + 2 min answer each

The technical part runs on CoderPad—a faithful reproduction of a real work scenario where you code and explain your thinking at the same time.

2. CoderPad Q1: Box Counting (Area of Box)

Problem

Given a 2D array of queries, each query is [row, col]. For each query, how many squares fit inside a row × col grid? A square has side length a with 1 ≤ a ≤ min(row, col).

Approach

Not hard—the key is to spot the pattern. Draw it out: a square of side a has (row - a + 1) × (col - a + 1) possible top-left positions. Sum over all valid side lengths.

def count_squares_for_queries(queries):
    results = []
    for row, col in queries:
        total = 0
        for a in range(1, min(row, col) + 1):
            total += (row - a + 1) * (col - a + 1)
        results.append(total)
    return results

Complexity O(Q · min(row, col)). If the input is large and the interviewer pushes, expand the sum into a closed form (a product/difference of arithmetic and square sums) to answer each query in O(1)—exactly the "optimize one more level" follow-up GS loves.

3. CoderPad Q2: Longest Subarray (sum at most k)

Problem

Given an array, find the length of the longest contiguous subarray whose sum is at most k.

Approach

Classic sliding window: expand the right pointer and accumulate; once the window sum exceeds k, shrink from the left, tracking the maximum length along the way.

def longest_subarray_with_sum_at_most_k(nums, k):
    left = 0
    cur = 0
    best = 0
    for right in range(len(nums)):
        cur += nums[right]
        while cur > k and left <= right:
            cur -= nums[left]
            left += 1
        best = max(best, right - left + 1)
    return best

O(n)—each element enters and leaves the window once. Watch the boundary: if the array contains negatives, monotonicity breaks and the sliding window fails; you'd switch to prefix sums + a monotonic deque. Interviewers often probe with "what if there are negatives?", so prepare that follow-up.

4. The 9 math multiple-choice (~40 minutes)

The prompts lean probability and linear algebra, and the pace is tight—just over 4 minutes per question. The directions seen in the debrief:

  1. Dice game: opponent scores on 4/5/6, you score on 1/2/3; the winner must reach 5+ and lead by 2. It's 5:5 now—what's your win probability? (absorbing-state Markov chain / tie recursion)
  2. Card probability: 5 random cards, 3 are aces—probability that 4 are aces (hypergeometric).
  3. Find the general term of a sequence given an integral.
  4. Differentiate an integral (Leibniz rule).
  5. A 3 × 3 linear system—decide whether the solution is CONSISTENT and UNIQUE (determinant / rank).
  6. Roll a die, reroll on a 1—probability of coming back to win after falling behind in a tie.
  7. Path / line integral.
  8. Three linear-algebra statements to judge true/false—eigenvalues, determinant, identity matrix.
  9. Sum of 100 Bernoulli variables with p=0.5—probability it is < 60 (normal approximation + continuity correction).

Tip: don't grind every MCQ to the end. For Q9, use the normal approximation N(50, 25): P(X<60) ≈ Φ((59.5-50)/5) ≈ Φ(1.9), done in 30 seconds; computing the exact binomial would blow the clock.

5. HireVue 6 prompts

GS's HireVue has 6 questions, 45s prep and 2 min to answer each—genuinely tight:

  1. Walk through your resume / background.
  2. Working with someone who wasn't completing their part—what did you do?
  3. A time you hit a high-challenge goal others thought you couldn't.
  4. A time you turned down a project or opportunity due to a conflict.
  5. You took an extra class and are overwhelmed; you're on an individual project that bans collaboration and a classmate wants to help—what do you do?
  6. How do you debug?

Use the STAR framework—prep 2–3 core project stories broken into Situation / Task / Action / Result so they reuse across most prompts.

6. Summary

The GS OA has a consistent flavor: coding questions that aren't brutally hard but demand think-aloud + one more optimization layer, math questions that reward fast approximation, and HireVue that rewards structured delivery. Train all three—don't just grind LeetCode.

FAQ

Q1: How many questions, and what platform, is the Goldman Sachs OA?

This round was 2 CoderPad-style coding questions + 9 math multiple-choice (~40 minutes) + a 6-prompt HireVue recording. Code while explaining your thinking.

Q2: How do you optimize the box-counting question?

The naive solution enumerates side lengths per query, O(Q·min(row,col)). On follow-up, expand the sum into a closed form (combining square and linear sums) to bring each query to O(1).

Q3: What's the trap in the longest-subarray question?

The sliding window assumes non-negative elements and a monotone window sum. Interviewers often ask "what about negatives?"—then switch to prefix sums + a monotonic deque. Prep that follow-up.

Q4: How do you speed up the math MCQs?

Approximate instead of grinding. Use a normal approximation with continuity correction for the Bernoulli sum, and absorbing-state recursion for the dice game—30 seconds each. For timed GS OA mocks and fast math-solving drills, the OA assist / OA live support path can tailor a plan.

Preparing for the Goldman Sachs OA?

The three GS OA blocks (coding + math + HireVue) each have their playbook. oavoservice offers full-process GS practice: timed CoderPad coding mocks with "optimize one more level" follow-up drills, fast-solving heuristics for the 9 math questions, and HireVue STAR templates, with question-type prediction by role line. Coaches include senior ex-big-tech engineers who know GS's scoring style, with end-to-end OA assist / OA live support.

Add WeChat Coding0201 now to get GS questions and mock practice.

Contact