Pinterest Interview Questions — Page 8

Pinterest·Software Engineer·Technical Phone Screen

May 2026

Pinterest phone screen for a software engineer role, pretty much one meaty graph problem the whole time. The question had enough moving parts that I left unsure if I'd covered everything they wanted.

Given a list of airline tickets as directed pairs [from, to], reconstruct an itinerary starting from a given airport that uses every ticket exactly once. If multiple valid itineraries exist, return the lexicographically smallest one. Walk through your data structures, time and space complexity, and explain how your solution handles cycles and cases where no valid itinerary can be formed.

“This is basically Hierholzer's algorithm dressed up in an airport costume.”

View Post

Pinterest·Software Engineer·Technical Phone Screen

May 2026

Pinterest SWE interview with two coding problems back to back. The questions were more design-flavored than pure leetcode grind, which I wasn't totally expecting.

Implement a LineReader class with a nextLine() method that reads complete lines from a chunked stream source, where each chunk can contain zero or more newlines and the chunks must be processed with O(L_max) additional memory.
Given a list of money transfer transactions between people, compute a list of payback transfers so that every person's net balance returns to zero. You don't need to minimize the number of payback transactions.

“This one took me a minute to get comfortable with.”

View Post

Pinterest·Machine Learning Engineer·Technical Phone Screen

May 2026

Pinterest ML engineer screen, all fundamentals. Five questions back to back, felt like a written exam more than a conversation. Nothing crazy hard but you need to actually know your stuff cold.

Define precision and recall for a binary classifier and explain how each maps to the confusion matrix.
Given specific confusion matrix values (TP, FP, TN, FN), compute precision and recall and explain which types of errors move each metric up or down.
What does a gradient represent mathematically, and how is it used in gradient descent?
Walk through the tradeoffs between bagging and boosting, including bias-variance and compute considerations.
Do a forward pass by hand through a small feedforward neural network: matrix multiply the inputs by the weights, add bias, apply an activation function, and give the final output.

“Pretty standard but I always fumble the wording under pressure.”

View Post

Pinterest·Machine Learning Engineer·Technical Phone Screen

May 2026

Pinterest ML engineer interview, technical phone screen that was basically a pure DSA session. Two parts to one problem, both on trees and linked lists. Nothing ML-specific came up at all, which surprised me.

Given a binary search tree, convert it in-place to a sorted circular doubly linked list where the left pointer acts as predecessor and the right pointer as successor. The smallest element's predecessor should wrap around to the largest. Solve in O(n) time with O(h) extra space.
Follow-up: given the sorted circular doubly linked list from the previous problem, write a function to insert a new value while keeping the list sorted. Your solution must correctly handle an empty list, inserting before the smallest element, inserting after the largest element, and duplicate values.

“The in-place constraint is where it gets annoying.”

View Post

Pinterest·Software Engineer·Technical Phone Screen

May 2026

Pinterest SWE interview that threw a group expense settlement problem at me, the kind that looks like a simple graph question until you realize there's actual algorithmic depth hiding underneath it.

Given a list of payments where each entry records who paid, how much, and on whose behalf (one or more people), compute the minimum number of transfers needed to settle all debts in the group and output the list of transfers as debtor, creditor, and amount.

“My first instinct was to just compute net balances and greedily match the biggest debtor to the biggest creditor using a max-heap.”

View Post

Pinterest·Machine Learning Engineer·Technical Phone Screen

May 2026

Pinterest ML engineer screen, rapid-fire fundamentals round. Three questions back to back with not much time to breathe between them. Felt like a quiz more than a conversation.

What is overfitting, and how do you explain it in terms of training versus validation performance?
What techniques would you use to prevent or reduce overfitting, and when would you reach for each one?
What's the difference between L1 and L2 regularization, covering their effects on weights, feature selection, and when you'd use each?

“Pretty standard opener but I fumbled the wording a bit.”

View Post

Pinterest·Machine Learning Engineer·Technical Phone Screen

May 2026

Pinterest ML Engineer interview that came down to a classic backtracking puzzle. Nothing flashy, just grind through the board and hope your constraint logic doesn't blow up.

Given a partially filled 9x9 Sudoku board, write a function to fill in the empty cells so the completed board is valid (each row, column, and 3x3 box contains digits 1 through 9 with no repeats).

“Backtracking was the obvious move but I spent probably too long trying to be clever about constraint propagation before just committing to the recursive approach.”

View Post

Pinterest·Data Scientist·Technical Phone Screen

May 2026

Pinterest data scientist coding screen, two parts back to back. Pretty much pure Python, no SQL, which I wasn't fully expecting. The problems themselves weren't brutal but the edge cases tripped me up a bit.

Given a list of pin engagement records and a category lookup map, find the category with the highest average time spent among video pins only. Null time values should be excluded, ties broken alphabetically by category name. Return the category name and rounded average. Must run in O(n).
Given a mapping of users to lists of pin IDs they interacted with (including duplicates), compute the average number of unique pin IDs per user. Users with empty lists count as zero. Return a float rounded to 2 decimal places, targeting O(total items) time and O(number of users) extra space.

“The None handling was the first thing I fumbled.”

View Post

Pinterest·Software Engineer·Technical Phone Screen

May 2026

Pinterest system design screen for a software engineer role. The whole thing was basically one big design problem about building a violation log analyzer, and they kept pushing deeper on every answer I gave.

Given an append-only list of violation events (each with an id, policy, and date), design an in-memory Violation Log Analyzer that can efficiently answer: which policies did a given id violate, which ids violated a given policy, and which ids violated any policy on a specific date.
If the event list is sorted by date, how would you use binary search to efficiently locate all events for a target date and return the corresponding ids? Walk through the algorithm and its complexity.
What are the trade-offs between query latency and memory usage in this design, and how does it scale as the number of events, distinct ids, and distinct policies grows?

“Started with the obvious brute force scan and they let me talk through it before asking about time complexity.”

View Post

Pinterest·Machine Learning Engineer·Onsite - System Design / Architecture

May 2026

Pinterest ML engineer interview focused entirely on a fraud detection system design question. Pretty deep dive, they wanted you to walk through the whole pipeline from problem framing to monitoring, not just slap a model on it and call it done.

Design an ML system to detect fraudulent transactions on an e-commerce platform. Walk through your full approach from problem framing to deployment and monitoring.

“This question is deceptively wide.”

View Post