RepostedAdobe logo

Adobe · OA

In process

Adobe New Grad OA: Unit Conversion and Code Review: 1 Experience

Software engineer · New GradBachelor's degreeReported Nov 19, 2026
Shared byjanebeingjane

A 2026 Adobe new grad OA in two parts: a LeetCode unit conversion question solved as a graph walk, then debugging a Django ticket confirm flow.

At a glance

  • Process: applied online → an online assessment (OA) in two parts: one coding question, then backend debugging and code review (report 01)
  • Asked recently: unit conversion over a graph of conversion rates, and fixing how a held ticket gets confirmed in a Django app with Redis (report 01)
  • Difficulty: the candidate rated it average and was put on the waitlist (report 01)
  • Practise first: Base Unit Conversions, Fix Confirming a Held Ticket

Coverage: One new graduate with a Bachelor's degree who applied online for a full-time general software engineering role. The report covers the OA only. It gives July to September 2026 as the interview period and went up on November 19, 2026.

What came up

One part was a graph traversal and the other was a debugging task about keeping a database, counters and Redis in step (report 01).

Topic Reports Last seen
OA coding: unit conversion as a weighted graph 1 of 1 Q3 2026
OA debugging and code review: confirming a held ticket 1 of 1 Q3 2026

Questions we found

Algorithms / Data Structures

Question Reports Last seen
Unit conversion (practice) 1 Q3 2026

Project Coding

Question Reports Last seen
Ticket confirmation debugging (practice) 1 Q3 2026

1 detailed candidate timeline, loops, and screens

01: New grad software engineer, two-part OA, waitlisted

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, average difficulty · Result: Waitlisted

OA part 1: unit conversion

  1. The candidate marks this part as a LeetCode question and files it under graph or tree traversal.
  2. The input is a list of conversions between units. From a starting unit, work out the matching value of every unit linked to it. If unit A converts to unit B with factor x and A has value v, then B has value v times x.
  3. The candidate calls it a very classic weighted graph traversal. Their approach treats the conversions as a weighted graph (undirected, or a directed tree), runs a DFS or BFS from the base unit, and passes the running product of factors down each edge. They give O(N) time and O(N) space, where N is the number of units.
  4. FastPrep's practice version is Base Unit Conversions. The candidate gives no input format or limits, so FastPrep uses the setup of LeetCode 3528 (Unit Conversion I): units numbered 0 to n − 1, unit 0 as the start, and answers taken modulo 10^9 + 7.

OA part 2: backend debugging and code review

  1. The task was ticket confirmation debugging. The candidate lists what it tested: Django, backend work, keeping state consistent, and Redis.
  2. In a ticket sales app, the confirm-purchase flow is broken and needs fixing. A ticket in the held state has to become confirmed, and five things must agree afterwards: the ticket's status in the database, the section's stock counts, the hold's expiry, the hold's lock in Redis, and whether the event is sold out.
  3. A confirm only goes through when the ticket is still held, its hold hasn't run out, and it belongs to the user asking. After a success, the status is confirmed, hold_expires_at is set to null, the section's held_count drops by one and its sold_count rises by one, the hold key is removed from Redis, and as the last step the event is checked for selling out.
  4. Each refusal has a set response: a 400 error when the ticket isn't held, another 400 when the hold has expired, and a 404 when the ticket can't be found or isn't this user's. The task fixes the message text for each.
  5. The candidate's notes: the task leans very hard on consistency across several layers (database, hold lifecycle, stock and Redis). The bug most likely to be missed is in the Redis helper, which builds the hold key but never actually deletes it. The order of the checks also matters, for example looking for a missing ticket before checking for an expired hold.
  6. FastPrep built Fix Confirming a Held Ticket from this report. It plants more bugs than the Redis one the candidate names.

Result

The candidate was put on the waitlist.


Common questions

What is on the Adobe new grad OA?

In report 01, two parts: a unit conversion question solved as a graph traversal, then debugging and reviewing a Django ticket confirmation flow.

What does the Adobe OA code review part ask?

In report 01, fixing the flow that turns a held ticket into a confirmed one so that the database, section counts, Redis hold key and sold-out status agree, with set errors for refused confirms.

Keep practicing

Sources: Anywhere useful interview info can be found.

0 total reactions. Your reactions: none.
0 comments

Comments

0

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