← Back to Directory

LinkedIn interview notes, page 4 of 32

Large Enterprises

Most of these LinkedIn write-ups come from Software Engineers (156), then Product Managers (66) and Data Scientists (37). The most reported rounds are the technical phone screen (169), the product sense round (62), the system design round (42) and the behavioral / leadership round (20).

316 interview notes · updated Oct 2026

LinkedIn·Software Engineer·Onsite - Coding / Algorithms

Apr 2026
LinkedIn SWE coding round with two parts: debugging a broken priority queue and then building a thread-safe concurrent queue for a web crawler scenario. The AI assistant was there but you were expected to talk through your thinking rather than just copy whatever it suggested, which honestly changes the dynamic a lot.
  • You're given a buggy priority queue implementation. Find and fix all the bugs, including heap property violations, comparator misuse, off-by-one errors in sift-up and sift-down, and edge cases for empty or single-element heaps.
  • Implement a thread-safe concurrent queue to back a web crawler, supporting multiple producer threads enqueuing URLs and multiple consumer threads dequeuing them, with proper blocking behavior, graceful shutdown, and no busy-waiting.

“The off-by-one in sift-down got me.”

View Post

LinkedIn·Data Scientist·Onsite - Product Sense / Strategy

Apr 2026
Data Scientist interview at LinkedIn with a pretty product-heavy feel. Three questions, all open-ended and estimation or framework-driven. No coding, just a lot of thinking out loud about users and data.
  • Segment LinkedIn's user base by identifying and describing the five most important user archetypes.
  • LinkedIn has around 500 million registered members. Walk through how you'd estimate the number of those who are sales professionals.
  • Build a data-driven framework to predict which LinkedIn members are likely to adopt a new product being promoted via email campaign.

“I went with job seekers, recruiters, sales people, content creators, and passive professionals.”

View Post

LinkedIn·Machine Learning Engineer·Technical Phone Screen

Apr 2026
LinkedIn ML engineer interview with a technical question on model evaluation. Pretty straightforward topic but there's more nuance to it than you'd expect if you haven't thought carefully about data leakage and holdout strategy.
  • How does cross-validation work in the context of logistic regression, and how does it differ from a standard train/test split?

“I went straight into k-fold and explained how you rotate the holdout fold each time to get a more stable estimate of generalization error.”

View Post

LinkedIn·Machine Learning Engineer·Technical Phone Screen

Apr 2026
Did a technical screen for an ML Engineer role at LinkedIn that had a geometry/math flavor I wasn't expecting. One problem, clean setup, but the edge case handling is where things get interesting.
  • Given a line in 2D defined by the equation ax + by + c = 0 and a list of points, compute the perpendicular distance from each point to the line and return the distances in the same order as the input. Make sure to handle the degenerate case where the line is invalid.

“The core formula isn't hard once you remember it: absolute value of (a*px + b*py + c) divided by sqrt(a squared plus b squared).”

View Post

LinkedIn·Data Scientist·Technical Phone Screen

Apr 2026
SQL-heavy technical screen for a Data Scientist role at LinkedIn. Three questions, all around the same two tables, each one a bit more annoying than the last. Nothing too exotic but the third one tripped me up more than I expected.
  • Using a members table and a video_posts table, count how many members uploaded their very first video on the same calendar date they joined the platform.
  • Compare total video uploads between US members and non-US members, where US is defined as country = 'usa'.
  • For each country, find the number of distinct members who have posted at least one video longer than 60 seconds.

“Straightforward once you see it.”

View Post

LinkedIn·Product Manager·Onsite - Product Sense / Strategy

Apr 2026
LinkedIn product sense question, pretty open-ended. Pick a product, say how you'd make it better. Sounds easy until you're actually sitting there trying to decide whether to talk about Jobs, Feed, or something less obvious.
  • Pick a LinkedIn product and walk through how you would improve it.

“I went with the Jobs tab because I use it constantly and figured familiarity would help.”

View Post

LinkedIn·Software Engineer·Technical Phone Screen

Apr 2026
LinkedIn SWE interview that came down to two graph problems, both island-related. Pretty focused session, no behavioral fluff, just code.
  • Given a 2D grid of '1's and '0's, count the number of islands (connected regions of land).
  • Given a 2D grid, count the number of distinct island shapes, where two islands are considered the same if one can be translated to match the other.

“Classic BFS/DFS problem and I knew it, but I still fumbled explaining the visited tracking out loud.”

View Post

LinkedIn·Machine Learning Engineer·Technical Phone Screen

Apr 2026
LinkedIn ML Engineer technical screen covering the classics: overfitting/underfitting and regularization. Nothing exotic, but the depth they expected on L1 vs L2 was more than I anticipated going in.
  • What is overfitting and underfitting, and how do you tell which one you're dealing with from your training and validation curves?
  • Walk me through the approaches you'd use to reduce overfitting in a model.
  • What's the difference between L1 and L2 regularization, and when would you choose one over the other?

“I knew the definitions cold but fumbled a bit explaining the curves part.”

View Post

LinkedIn·Software Engineer·Technical Phone Screen

Apr 2026
LinkedIn SWE interview with a data structure design problem that looked clean on paper but had some tricky edge cases around duplicates. Not a bad experience, just required more careful thinking than I expected.
  • Design a data structure that supports add, delete, and getRandom operations all in average O(1) time, where duplicates are allowed and getRandom should return elements with probability proportional to their frequency.

“The random part is what gets you.”

View Post

LinkedIn·Product Manager·Onsite - Product Sense / Strategy

Apr 2026
LinkedIn PM interview with a single product design question. Pretty stripped down, no behavioral warmup, just jumped straight into it.
  • How would you design the YouTube landing page?

“I went straight into user segmentation which felt right, but I spent way too long on it and barely had time to actually talk about the page design itself.”

View Post