Three rounds at Adobe, before anyone asked about AI
| Company | Adobe |
|---|---|
| Team | Workfront, Reviews & Approvals |
| Role | Software Engineer Intern |
| Location | Salt Lake City, UT. Virtual loop |
| Date | November 2024 |
| How it started | Cold application |
| Rounds | 3, plus an offer call |
| Outcome | Offer. Accepted |
Adobe has just opened its SWE internship and new grad roles, so the internship loop is worth writing up now. One thing to set straight at the top: this loop ran in November 2024. It is two cycles old, and the interview landscape has moved since. Read it as a record, not as this year’s answer key.
It is also the third interview post here and the first that ends in an offer. The other two are four rounds at Amazon and five rounds at DeepInfra, both rejections.
The loop
| Round | What it was | Format | Self Rating |
|---|---|---|---|
| 1 | Online assessment | Three LeetCode mediums on HackerRank | |
| 2 | Problem solving, live | One string manipulation problem. Medium | Strong Hire |
| 3 | Hiring manager fit | Behavioural. Background and the team | Strong Hire |
| 4 | Not a round | HR call with the offer |
I work at Adobe now, so I am describing the shape of the loop and not the questions inside it. Which round tests what is useful to know and is on Glassdoor already. The specific problem I was given is my employer’s, not mine to hand over.
Nobody asked about AI
Three rounds, of which exactly one was a live algorithm interview. The assessment in front of it was three timed mediums with no interviewer in the room, and the round behind it was a conversation with the person who would manage me.
Nothing in any of it touched AI. Not how I use models, not how I review what they write, not where I would refuse to. In November 2024 that was unremarkable. Writing it down in late 2026 it is the single most dated thing about the loop, and I would expect that gap to be where a 2026 loop spends its time instead.
One algorithm round
Set this against the other end of the range. DeepInfra ran five rounds, three of them live problem solving, pitched by a team of ICPC and IOI medallists, and ended on a plane problem I did not solve. Amazon ran four rounds at its Seattle HQ, a flight each way, and then nobody called back.
The shortest loop I have sat produced the only offer that turned into a career. The two longest produced nothing. I would like to report a lesson in that and I do not have one: the sample is small, and the variable I most want to isolate, how hard the loop was, is not the variable that decided anything.
What it converted into
The internship became a return offer, and the return offer was for Software Engineer II. Same product and the same team, Workfront Reviews and Approvals, which is where I still am. No second loop: no full-time application, no repeat of the algorithm round, no onsite.
Which means the entirety of my full-time interview experience at Adobe is the three rounds above, sat in November 2024, for an internship.
That is also why there is no separate Adobe entry in my interview log for the full-time job. One loop, two offers. Counting the conversion as a loop would put an offer on that page with no rounds behind it.
If you are applying this cycle
Caveat first, again: this is a 2024 loop. Assume the structure has drifted and that the AI-shaped gap above has been filled in by now.
What probably still holds is where the filter sits. The online assessment is three timed mediums and it removes most applicants before a human has looked at anything. Everything after it was lighter than the assessment implied.
That cuts both ways. A short loop is less to prepare for, and it is also fewer chances to recover from one bad round.
It was not luck
A three-round loop with one medium in it sounds like getting away with something, and the honest version is the opposite. The reason one live algorithm round was enough is that it was not really being answered in that hour. It was being answered by every rated contest behind it, which you can see on the front page.
Years of problem solving do not pay out on a schedule. They compound quietly and then settle up in a single afternoon, and from the outside that afternoon looks like a lucky draw.
"Luck" is usually just unnoticed hard work.
🥹 If this helped, a like means a lot.