← Back to blog Goldman CoderPad VO: Subarrays + Excel Column + STAR
VO Goldman Sachs

Goldman CoderPad VO: Subarrays + Excel Column + STAR

2026-06-12

As fintech moves faster and faster, CoderPad has become Goldman Sachs's go-to online coding platform—clean interface, strong real-time collaboration, and a clear read on your "code-and-explain" skill. This piece is organized from an oavoservice student's GS CoderPad VO debrief, walking through this round's algorithm question, programming question, and behavioral, plus how the VO live support / VO interview assist path plugs in.

1. Pre-interview preparation

CoderPad interviews usually run 1–1.5 hours and get into coding quickly. Questions are mostly above-average algorithm problems, supplemented by data-structure design and code optimization, with some rounds tied to financial business scenarios. Interviewers care about problem-solving logic, code quality, and communication, not just right/wrong.

2. Algorithm: find all non-repeating subarrays summing to target

Problem

Given an integer array nums, find all non-repeating contiguous subarrays whose sum equals target.

For example, nums = [1, 2, -1, 3, -2, 2], target = 3, the output should include [1, 2], [3], [1, 2, -1, 3], etc.

Approach

The array contains negatives, so a sliding window won't work (monotonicity is broken). The right approach is prefix sums + a hash map:

def subarrays_with_sum(nums, target):
    res = []
    prefix = 0
    # prefix sum -> all indices (open-interval left ends) for that prefix sum
    seen = {0: [-1]}
    for r, x in enumerate(nums):
        prefix += x
        need = prefix - target
        if need in seen:
            for l in seen[need]:
                res.append(nums[l + 1 : r + 1])
        seen.setdefault(prefix, []).append(r)
    return res

In the interview, explain the approach before writing: why a sliding window fails, how prefix sums turn a double loop into linear time, and what the hash map stores. GS values boundary handling (empty array, target=0, all-negative), so narrate it before you code.

3. Programming: Excel column number to title

Problem

Given an Excel column number columnNumber, e.g. 27, return the corresponding title (27 → "AA").

Approach

It's essentially base-26 conversion, but 1-indexed with no zero (A=1, not 0). So before each modulo you offset by -1, map the remainder to A~Z, then integer-divide and continue with the higher digits.

def column_title(n: int) -> str:
    chars = []
    while n > 0:
        n -= 1                      # key: 1-indexed, there is no 0
        chars.append(chr(n % 26 + ord('A')))
        n //= 26
    return ''.join(reversed(chars))

Complexity O(log₂₆ n). The easy mistake is that n -= 1: drop it and 26 wrongly becomes "AZ" instead of "Z". Comment the key step to show clean programming logic.

4. Behavioral (STAR framework)

GS's Behavioral digs into details, so tell complete stories with STAR (Situation / Task / Action / Result):

5. Summary

The two GS CoderPad VO questions are typical: the algorithm question tests the prefix-sum + hash combo (with negatives deliberately added to rule out a sliding window), and the programming question tests boundary details of base conversion. Both require explaining the idea first, then writing clean, commented code. The behavioral uses STAR to tell complete stories.

FAQ

Q1: How long is the GS CoderPad VO, and what does it test?

Usually 1–1.5 hours, getting into coding quickly. Mostly above-average algorithm problems, plus data-structure design and code optimization, some tied to financial scenarios. It looks at problem-solving logic, code quality, and communication.

Q2: Why can't the target-sum subarray question use a sliding window?

The array contains negatives, so the window sum is no longer monotone and a sliding window fails. Use prefix sums + a hash map to bring "does prefix - target exist" to O(1) and the whole thing to O(n).

Q3: Where is the Excel column title question most error-prone?

Base-26 is 1-indexed with no zero, so you must n -= 1 before each modulo; drop it and 26 wrongly becomes "AZ".

Q4: How do you prepare the behavioral?

Use STAR to break 2–3 core stories into Situation/Task/Action/Result, covering "high-challenge goal" and "team disagreement". For timed GS CoderPad mocks and live think-aloud practice, the VO live support / VO interview assist path can tailor a plan.

Preparing for the Goldman Sachs CoderPad VO?

GS CoderPad is about "code-and-explain + stable code." oavoservice offers full-process GS practice: timed mocks on prefix-sum/hash combo questions and base-conversion boundary questions, live think-aloud practice on CoderPad, STAR behavioral polishing, and question-type prediction by role line. Coaches include senior ex-big-tech engineers who know GS's scoring style, with end-to-end VO live support / VO interview assist.

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

Contact