Stripe New Grad SWE OA and Phone Screen: 2 Experiences

Shared by janebeingjane ·
Company
Stripe
Reports
2 interview reports
Level
New Grad
Stages
OA → Phone screen

Stripe new grad SWE reports from 2026: a phone screen on the known fee calculator question, and a two-question OA with LLM retry logic.

At a glance

Coverage: Two new-grad candidates for full-time general software engineering roles, with reports posted on November 18 and 19, 2026. Report 01 is a phone screen from October to December 2026. Report 02 is an OA; its interview period is given as July to September 2026.

What came up

The two reports share no question. Report 02 describes both OA tasks in detail; report 01 names its question only.

Topic Reports Last seen
Fee calculation 1 of 2 Nov 2026 (report date)
Arrays: jumping a fixed distance 1 of 2 Q3 2026
Retries and error handling around an API 1 of 2 Q3 2026

Questions we found

Algorithms / Data Structures

Question Reports Last seen
Transaction Fee Calculator (Parts 1–2) 1 Nov 2026 (report date)
Maximum Game Score with Fixed Forward Jumps (same task as LeetCode 3147) 1 Q3 2026
Reliable LLM API calls (practice) 1 Q3 2026

2 detailed candidate timelines, loops, and screens

01: Phone screen with the fee calculator, passed

Report date: November 18, 2026 · Interview period: 2026 Q4 · Role: Software engineer, new grad · Degree listed: Master's · Channel: Referral · Candidate's rating: Neutral experience, average difficulty · Result: Passed

Phone screen

  1. The candidate got the transaction fee calculator and calls it a known original question. The report doesn't say which parts came up. FastPrep's base version is Transaction Fee Calculator (Parts 1–2).

Result

The candidate passed.


02: Two-question OA, not finished

Report date: November 19, 2026 · Interview period: 2026 Q3 · Role: Software engineer, new grad · Degree listed: Bachelor's · Channel: Applied online · Candidate's rating: Neutral experience, hard · Result: Did not pass

OA, question 1: maximum game score

  1. The input is an integer array game_val of length n and an integer k. The player may start at any index. From index i, the player jumps forward exactly k places to i + k, as long as that index is still inside the array.
  2. The score is the sum of every value the player lands on, including the start. The goal is the highest score over all possible starting points.
  3. Example: for game_val = [2, -3, 4, 6, 1] and k = 2, the answer is 7. Start at index 0 and visit indices 0, 2 and 4, for 2 + 4 + 1.
  4. The candidate tagged it as a LeetCode problem. It is the same task as LeetCode 3147 (Taking Maximum Energy From the Mystic Dungeon). FastPrep's version, Maximum Game Score with Fixed Forward Jumps, keeps jumping until the next jump would leave the array, so the start is the only choice.

OA, question 2: reliable LLM API calls

  1. Write a function reliable_call that talks to an LLM API and copes with failures. The candidate labels it low-level design and describes it as a mix of system design, object-oriented code and exception handling.
  2. Two helpers are given. client.complete(messages) sends the messages and may raise ClientError. validate_fn(response) checks the answer and may return False or raise ValidationError.
  3. The caller's conversation must stay untouched: don't change the original list, and preferably work on a deep copy.
  4. On a ClientError, add the CLIENT_ERROR_MESSAGE to the messages and try again.
  5. On a failed validation, add the bad response as an assistant message, then a user message that corrects it, and try again. When validate_fn raises ValidationError, the correction says the response was invalid, gives the reason and asks again for valid JSON.
  6. Client errors and validation failures each have their own retry count, and the two must not share a budget. A budget of 0 means no retries: one call only.
  7. Return the response as soon as it passes validation. When a budget runs out, raise LLMRetryError carrying last_response, attempts, all_responses and failure_reason. Here attempts counts every API call, and all_responses holds every response that failed validation. When client errors run out, the reason given is client_error.
  8. FastPrep's practice version is Reliable LLM Call with Independent Retry Budgets. It replaces the live client and validator with a scripted list of outcomes.

Result

The candidate didn't finish the OA and did not pass.


Common questions

What does the Stripe new grad phone screen ask?

In report 01, the transaction fee calculator, which the candidate recognised as a known question. The candidate passed.

What is on the Stripe new grad OA?

Report 02 had two questions: a fixed-jump array score (the same task as LeetCode 3147) and a function that retries LLM API calls on two separate budgets.

Is the Stripe new grad OA hard?

The one OA candidate rated it hard and didn't finish (report 02).

Other Stripe cohort: the experienced-hire reports from the same days, two OAs and an onsite, are in Stripe Experienced SWE OA and Onsite Questions: 3 Experiences.

Keep practicing

Sources: Anywhere useful interview info can be found.

0 total reactions. Your reactions: none.

Comments

0

No comments yet — be the first to share what you know.