← Back to Directory

Meta interview notes, page 128 of 238

FAANG / Big Tech

Most of these Meta write-ups come from Software Engineers (1,194), then Data Scientists (489) and Product Managers (453). The most reported rounds are the technical phone screen (1,190), the product sense round (466), the system design round (255) and the behavioral / leadership round (205).

2377 interview notes · updated Jul 2026

Meta·Machine Learning Engineer·Technical Phone Screen

May 2026
Meta ML engineer coding round, just the one question about string manipulation. Pretty standard but the edge cases will get you if you're not careful.
  • Given a string containing parentheses, write a function that removes the fewest characters possible to make the parentheses balanced.

“My first instinct was to reach for a stack and just track unmatched opens and closes separately.”

View Post

Meta·Software Engineer·Onsite - Behavioral / Leadership

May 2026
Behavioral round at Meta, one question about pushing through a long-term challenge. Pretty standard but the multi-month framing tripped me up a bit.
  • Describe a time you had to keep going through a challenge that stretched over several months.

“I had a decent story ready but the 'several months' part made me second-guess whether my example was long enough.”

View Post

Meta·Product Manager·Onsite - Product Sense / Strategy

May 2026
Meta product sense round, one question about recommending similar videos on YouTube. Not much else to say, it was a pretty focused session.
  • How would you design a system to recommend similar videos on YouTube?

“I went straight to user intent and started talking about watch history, engagement signals, that kind of thing.”

View Post

Meta·Software Engineer·Technical Phone Screen

May 2026
Meta SWE coding round, tree traversal question. Pretty standard stuff but the dual-view twist made me think twice about how I was structuring my BFS.
  • Given the root of a binary tree, return both the left side view and the right side view (the leftmost and rightmost node at each level, top to bottom).

“I jumped straight to BFS which was fine, but I got a little tangled trying to return both views in one pass.”

View Post

Meta·Software Engineer·Onsite - System Design / Architecture

May 2026
Meta system design round for a software engineer role. The whole session was basically one big question about location-based search, but it branched into a lot of sub-topics fast.
  • Design a location-based search service that takes a latitude, longitude, radius, and K as input and returns the top K nearby locations.
  • How would you index and query geospatial data efficiently? Walk through options like geohashing, quadtrees, or other approaches.
  • What are the tradeoffs between handling static versus dynamic location data in your design?
  • How would you shard this system, and would you consider any hybrid sharding strategies?
  • Do a rough back-of-envelope estimate for memory usage and query throughput at global scale.

“The question sounds straightforward until you realize 'location' is intentionally vague.”

View Post

Meta·Data Scientist·Technical Phone Screen

May 2026
Meta data scientist technical screen, heavy on stats and experimentation theory. The question was essentially one big multi-part problem about variance estimation and A/B testing design for a ratio metric, and it went pretty deep pretty fast.
  • Your team's KPI is ARPU, defined as Revenue divided by Active Users, computed per experiment bucket. Given that the correlation between Revenue and Active Users is 0.6, use the delta method to derive an approximate variance for ARPU, including the covariance term. How would you estimate the required inputs from sample data?
  • Compare two analysis approaches for this experiment: testing ARPU directly versus modeling log-Revenue with log-Active Users as a covariate, something like an ANCOVA setup. What are the tradeoffs?
  • When is a standard t-test acceptable for testing ARPU, and when would you switch to a bootstrap? Also, how do you handle heavy-tailed revenue distributions, for example through winsorization or robust estimators?

“This is where I started sweating.”

View Post

Meta·Software Engineer·Technical Phone Screen

May 2026
Meta SWE coding screen, one question about filtering and reconstructing digits from an integer. Pretty short session, nothing too wild.
  • Given an integer, return the smallest nonnegative integer that can be formed using only its odd digits. If none of the digits are odd, return None. For example, 62315 becomes 135, 260 returns None, and -25 returns 5.

“The negative number case tripped me up briefly.”

View Post

Meta·Software Engineer·Technical Phone Screen

May 2026
Coding round for a Research Scientist role at Meta. Just one algorithmic problem but it had enough moving parts to keep me busy the whole session.
  • You have two sorted lists of departure times representing flights from A to B and from B to A. Given a current time, simulate completing full round trips: find the earliest departure no earlier than the current time in each direction, update the current time accordingly, and repeat for a given number of missions.

“The sorted lists made me think binary search immediately, which was the right call for finding the next valid departure.”

View Post

Meta·Data Scientist·Technical Phone Screen

May 2026
Meta Data Scientist interview with a heavy stats and probability focus. The core problem was a multi-part fake account detection question that went pretty deep into Bayes, binomial distributions, and then got messier when they introduced signal correlation at the end.
  • You have a fake account classifier that uses five independent signals. Each signal fires with probability 0.8 for fake accounts and 0.05 for real ones. For a threshold of k=2 (flag if at least 2 signals fire), compute the probability of flagging a fake account and the probability of flagging a real account using the binomial distribution.
  • Given a base rate of 1.5% fake accounts and 5 million accounts scanned per day, use Bayes' theorem to find P(fake | flagged) and the expected number of flagged accounts. Is an 80,000-per-day manual review queue sufficient?
  • Find the smallest threshold k such that the expected daily flagged volume stays within 80,000 plus or minus 5%, while also maximizing P(fake | flagged). Walk through the trade-off between precision and recall.
  • If the five signals are not actually independent and have a pairwise correlation of 0.2 within each class, how does that change your analysis and what assumptions break down?

“Binomial CDF complement, nothing exotic.”

View Post

Meta·Software Engineer·Onsite - Coding / Algorithms

May 2026
Meta SWE coding round, got a binary array range query problem. Pretty classic prefix sum setup but they pushed on complexity and edge cases more than I expected.
  • Given a large immutable binary array, design a data structure that preprocesses it once and then answers range queries for the count of 1s in any inclusive interval [l, r] as efficiently as possible.

“My first instinct was just to scan the range each query, which is obviously too slow at 10^7 elements and 10^5+ queries.”

View Post