LinkedIn · Phone screen → Onsite
OfferLinkedIn Senior MLE Interview Questions: 1 Experience
A 2026 LinkedIn senior MLE loop that ended in an offer: LC 21 and ML basics by phone, then five onsite rounds, the hardest a squared-distance DP.
At a glance
- Process: recruiter contact (May and late June 2026), a phone screen in August, five 1-hour onsite rounds in early September, and an offer in mid-September (report 01)
- Asked recently: LC 21; a squared-distance version of LC 1478 with a mean-versus-median proof; a recommendation system design; ML knowledge questions (report 01)
- Difficulty: the candidate rated the loop hard, but only the onsite coding round with the mailbox variant felt hard; the other rounds went fine
- Practise first: Allocate Mailboxes with Squared Distance, Merge Two Sorted Lists, Design a Personalized Recommendation System
Coverage: One experienced machine learning engineer changing jobs, with a Master's degree, who was brought in by an agency recruiter. The job title was Software Engineer - AI, which the candidate describes as a senior MLE role. The video interviews ran from August to early September 2026, and the report went up on November 19, 2026. It gives no location.
What came up
In this one loop, coding came up in the phone screen and in two onsite rounds, and ML knowledge in the phone screen and one onsite round.
| Topic | Reports | Last seen |
|---|---|---|
| Coding: DP and linked lists | 1 of 1 | Sep 2026 |
| ML knowledge questions | 1 of 1 | Sep 2026 |
| ML system design: recommendations | 1 of 1 | Sep 2026 |
| Behavioral and resume projects | 1 of 1 | Sep 2026 |
Questions we found
Algorithms / Data Structures
| Question | Reports | Last seen |
|---|---|---|
| Allocate mailboxes with squared distance (LC 1478 variant) | 1 | Sep 2026 |
| Merge two sorted lists (LC 21) | 1 | Aug 2026 |
System Design
| Question | Reports | Last seen |
|---|---|---|
| Recommendation system | 1 | Sep 2026 |
ML knowledge
| Question | Reports | Last seen |
|---|---|---|
| Neural networks, in depth | 1 | Sep 2026 |
| Decision trees | 1 | Sep 2026 |
| Overfitting and underfitting | 1 | Aug 2026 |
| Dropout | 1 | Aug 2026 |
| Regularization | 1 | Aug 2026 |
1 detailed candidate timeline, loops, and screens
01: Senior MLE (Software Engineer - AI), full loop to an offer
Interview period: August to September 2026 · Report date: November 19, 2026 · Role: Software Engineer - AI (machine learning engineer), senior · Degree listed: Master's · Channel: Recruiting agency · Format: Video · Candidate's rating: Positive experience, hard · Result: Offer
Recruiter contact (May to June 2026)
- A recruiter first reached out in May 2026. After a short chat, the recruiter said a position would open in July and promised to get back in touch then.
- At the end of June, the recruiter came back and offered to set up interviews.
Phone screen (August 2026)
- Coding: LC 21, merging two sorted linked lists into one sorted list. FastPrep's practice version is Merge Two Sorted Lists.
- ML basics: a deep dive into neural networks, then a sweep through most of the ideas in an intro machine learning course (the candidate compared it to Andrew Ng's ML course). Overfitting and underfitting, dropout and regularization were among them.
Onsite (early September 2026): five rounds of 1 hour each, on one day or split over two
- AI coding round. LinkedIn had not yet moved this round to coding with AI tools, so it was a coding question built on an AI idea. The task was LC 1478 (Allocate Mailboxes) made harder. Each house's cost is its squared distance to the nearest mailbox (L2) instead of the plain distance (L1), and the goal is the smallest mean squared distance. The candidate said a plain DP is not enough here; the best solution adds running totals (prefix sums) of the positions and of their squares. A second hard part was proving to the interviewer why, with squared distance, a group's mailbox belongs at the mean of its houses and not the median. The follow-up asked for something simpler if the answer didn't have to be optimal, and the candidate suggested k-means clustering. The candidate says they could only pass because they had read about this problem in an earlier candidate's report. FastPrep's practice version is Allocate Mailboxes with Squared Distance. It asks for the total squared distance instead of the mean; the best placement is the same.
- Coding round. An easy-to-medium question the interviewer had written. The candidate no longer remembers it.
- System design round. A recommendation system: retrieval of candidate items, ranking, features, training samples, and offline and online metrics. The interviewer moved from topic to topic freely and also asked about the resume, since the candidate comes from a recommendations background. FastPrep's practice version is Design a Personalized Recommendation System.
- Behavioral and hiring manager round. About half behavioral questions and half questions about projects on the resume. The report names no single question.
- ML knowledge round. Decision trees and neural networks. Then the interviewer shared a screen showing LinkedIn's own list of ML questions and went through about five of them one by one. To the candidate, it felt like a university exam.
Result
- Only the AI coding round was hard; the others went fine.
- The offer came in mid-September 2026.
Common questions
What is the LinkedIn MLE AI coding round like?
In one 2026 senior loop, it was a coding question on an AI idea with no AI tools: a squared-distance version of LC 1478, a proof that the mean beats the median, and a k-means follow-up (report 01).
How many onsite rounds are there for a LinkedIn senior MLE?
One candidate had five 1-hour rounds: AI coding, coding, ML system design, behavioral with the hiring manager, and ML knowledge, on one day or two (report 01).
How long does the LinkedIn MLE hiring process take?
For one candidate it took about four months from the first recruiter message in May 2026 to the offer in mid-September; the interviews themselves fell in August and early September (report 01).
What to practise
- Allocate Mailboxes with Squared Distance. The one round the candidate in report 01 called hard, September 2026.
- Merge Two Sorted Lists. The phone screen coding in report 01, August 2026.
- Design a Personalized Recommendation System. The ML system design round in report 01, September 2026.
Keep practicing
- Practice questions: Allocate Mailboxes with Squared Distance and Merge Two Sorted Lists (report 01).
- Practice design: Design a Personalized Recommendation System (report 01).
- More LinkedIn coding questions: the LinkedIn coding question bank. It needs sign-in.
- More LinkedIn designs: the LinkedIn system design bank. It needs sign-in.
- LinkedIn behavioral questions: the LinkedIn behavioral bank, for the hiring manager round. It needs sign-in.
- Interview overview: the LinkedIn interview questions page.
- Hiring insights: the LinkedIn hiring insights page.
- More firsthand reports: browse the Experience feed.
Sources: Anywhere useful interview info can be found.
Comments
0No comments yet — be the first to share what you know.