LinkedIn · Onsite
RejectedLinkedIn Experienced MLE Onsite: 1 Experience
A LinkedIn experienced MLE onsite in late 2026 had five rounds, from probability-heavy ML fundamentals and video recommendation to AI-assisted training.
At a glance
- Process: a five-round onsite: modeling and fundamentals, behavioral, ML system design, AI-assisted coding, and data coding (report 01).
- Asked recently: probability, stochastic processes, normal and Poisson distributions; a video recommendation design; training a model on a dataset with Claude allowed; a maze where the mouse can jump back to visited cells (report 01, report date Nov 2026).
- Difficulty: rated average; the fundamentals felt like a trading interview, and the candidate used BFS on the maze without knowing A*. Not selected.
- Practise first: Explore an Unknown Grid and Find a Path and Design a Personalized Recommendation System.
Coverage: One experienced machine learning engineer with a Master's degree, changing jobs, who came to LinkedIn through a recruiter. The report covers a five-round onsite. It gives October to December 2026 as the interview period and went up on November 13, 2026.
What came up
The five rounds covered ML fundamentals, behavioral questions, ML system design, AI-assisted model training and a data coding question. The fundamentals round leaned so hard on probability that the candidate felt it could have been a trading interview.
| Topic | Reports | Last seen |
|---|---|---|
| ML fundamentals: probability, stochastic processes, normal and Poisson distributions | 1 of 1 | Nov 2026 (report date) |
| ML system design: video recommendation | 1 of 1 | Nov 2026 (report date) |
| AI-assisted coding: train a model on a given dataset | 1 of 1 | Nov 2026 (report date) |
| Data coding: a maze where the mouse can jump back to visited cells | 1 of 1 | Nov 2026 (report date) |
| Behavioral round (no questions named) | 1 of 1 | Nov 2026 (report date) |
Questions we found
Algorithms / Data Structures
| Question | Reports | Last seen |
|---|---|---|
| Mouse in a maze that may jump to any cell it has already visited; find the way through (Explore an Unknown Grid and Find a Path) | 1 | Nov 2026 (report date) |
ML System Design
| Question | Reports | Last seen |
|---|---|---|
| Design a video recommendation system (practice; the practice version is a general recommender, not video-specific) | 1 | Nov 2026 (report date) |
AI Coding
| Question | Reports | Last seen |
|---|---|---|
| Train a model on a given dataset, in several parts, with an AI assistant allowed | 1 | Nov 2026 (report date) |
1 detailed candidate timeline
01: Experienced MLE, five-round onsite
Interview period: 2026 Q4 · Report date: November 13, 2026 · Role: Machine learning engineer · Degree listed: Master's · Channel: Recruiter · Candidate's rating: Neutral, average difficulty · Result: Not selected
The report lists the five rounds as below. It doesn't say that this was the order on the day.
- Modeling and fundamentals. Lots of standard knowledge questions, plus probability and stochastic processes, including the normal and Poisson distributions. The candidate says that without context they would have guessed it was a trader interview.
- Behavioral. The report doesn't name the questions.
- ML design: video recommendation. No further detail is given. The closest FastPrep practice version is Design a Personalized Recommendation System, which covers a general user-item recommender rather than video specifically.
- AI-assisted coding. The candidate got a dataset and had to train a model, with Claude allowed, across several parts. The report doesn't describe the dataset or the parts, so this round has no practice version.
- Data coding: mouse in a maze. A mouse has to find its way through a maze. It can teleport to a position, but only one it has already been to, a rule the candidate found a bit strange. Not knowing A*, the candidate pushed through with BFS. Practice version: Explore an Unknown Grid and Find a Path, where a robot explores an unknown grid, may relocate to any cell it has already explored, and returns a shortest path.
Result
A week later the candidate heard they had not passed.
Common questions
What rounds are in the LinkedIn MLE onsite?
Five in this report: modeling and fundamentals, behavioral, ML system design (video recommendation), AI-assisted model training, and a data coding maze problem (report 01).
What does the LinkedIn MLE fundamentals round ask?
Standard ML knowledge plus a lot of probability and stochastic processes, including normal and Poisson distributions (report 01).
Can you use AI in the LinkedIn MLE coding round?
Yes, Claude was allowed in the model-training round (report 01).
Keep practicing
- Practice questions from this report: Explore an Unknown Grid and Find a Path (data coding) and Design a Personalized Recommendation System (ML design).
- More LinkedIn questions: the LinkedIn system design bank and the LinkedIn question bank, with every practice format.
- Everything LinkedIn in one place: the LinkedIn interview page and LinkedIn hiring insights.
- More firsthand reports: the Experience feed.
Sources: Anywhere useful interview info can be found.
Comments
0No comments yet — be the first to share what you know.