← Back to Directory

Meta interview notes, page 235 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·Data Scientist·Technical Phone Screen

Jul 2026
Meta data scientist interview with a probability question that sounds deceptively simple but requires you to actually think through Bayes properly. One question, pretty focused.
  • Lazy reviewers (20% of all reviewers) always leave positive feedback. Careful reviewers (the other 80%) leave positive feedback 60% of the time. Given that a review is negative, what is the probability it came from a lazy reviewer?

“This one tripped me up for a second because the answer feels like it should be zero (lazy reviewers never go negative), and it basically is.”

View Post

Meta·Data Scientist·Onsite - Product Sense / Strategy

Jul 2026
A product sense question dressed up as a Data Scientist interview at Meta, which wasn't exactly what I expected. The whole thing revolved around a Google Maps Group feature scenario and required senior-level PM-style thinking, not the SQL or experimentation stuff I'd been grinding.
  • A PM tells you the team wants to build a new Google Maps feature to increase Group page engagement. Walk through how you'd clarify scope, define users, set measurable goals aligned to company mission, propose success metrics with guardrails, and structure a 60-minute stakeholder alignment conversation.

“This was basically a full product strategy case and I kept waiting for it to pivot into something more analytical.”

View Post

Meta·Data Scientist·Technical Phone Screen

Jul 2026
Meta DS interview focused on the Circle group feature, basically a product analytics deep-dive that mixed metric design, experiment planning, and chart interpretation all in one session. Trickier than it sounds because the questions kept building on each other.
  • How would you define success metrics for the Circle feature compared to regular posts?
  • Design an experiment to evaluate the Circle feature under two scenarios: one where engineering resources are limited and one where they're plentiful. What trade-offs would you weigh in each case?
  • You're shown three line charts plotting total comments divided by total posts over time, one each for Circle, business pages, and friends posts. Can you directly compare these lines? What do the charts tell you, and what hypotheses would you form?

“I went straight to engagement metrics and the interviewer kind of waited for me to say more.”

View Post

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

Jul 2026
Brutal system design round at Meta for an MLE role. The question was basically 'design a full recommendation system from scratch' and they meant it, every layer, no shortcuts.
  • Design an end-to-end large-scale recommendation system for a consumer app's home feed, covering candidate generation, feature stores, real-time signal ingestion, ranking, re-ranking, and online exploration strategies like multi-armed bandits.
  • How would you handle cold start for both new users and new items in a recommendation system?
  • What are your strategies for detecting and mitigating feedback loops and bias in a recommendation system?
  • How would you design an A/B testing framework for a recommendation system where the treatment and control groups can influence each other through shared item pools?
  • What latency and throughput targets would you set for a home feed recommendation system, and how would your architecture enforce them?
  • How would you handle data privacy requirements within a recommendation system, and what changes to the architecture does that introduce?
  • What is your fallback strategy when parts of the recommendation system go down in production?

“I started with candidate generation and worked forward, which felt logical but in hindsight I think they wanted me to anchor on scale constraints first.”

View Post

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

Jul 2026
Meta ML Engineer system design round focused entirely on building a real-time nearby place recommendation system. Pretty comprehensive scope, they wanted the full stack from metrics down to A/B testing, so it was a lot to cover in one session.
  • Design a real-time system that recommends nearby places to users. Walk through your objectives and success metrics, data sources, candidate generation, feature engineering, ranking model, cold-start handling, exploration vs exploitation, spam filtering, privacy constraints, latency requirements, and an A/B testing plan.

“This is basically a full ML system design in one question.”

View Post

Meta·Machine Learning Engineer·Technical Phone Screen

Jul 2026
Meta ML engineer coding round, basically one meaty array problem with a follow-up that I did not see coming. The core question was fine but the streaming extension is what really separated people I think.
  • Given an unsorted array of up to a million integers and an integer k, return the k smallest elements in ascending order. Your solution should run in O(n log k) time and use O(k) extra space using a heap. How do you handle duplicates and negative numbers?
  • Follow-up: how would you redesign this for a streaming scenario where integers arrive one at a time and you need to support a getSmallestK() call at any point?

“I jumped straight to a max-heap of size k and walked through it, which was the right call.”

View Post

Meta·Data Scientist·Technical Phone Screen

Jul 2026
Meta DS interview with a product analytics focus, this one centered on push notifications for a travel app. The question had a few layers and I don't think I fully nailed the interference piece.
  • For a new push notification feature in a travel recommendation app, what primary success metrics would you track, and which guardrail metrics would you include to catch negative effects like uninstalls or unsubscribes?
  • The app has strong network effects because users share itineraries with each other. How would you design an A/B test for the push notification feature to reduce interference between groups, and what would your unit of randomization be?

“I went straight to engagement stuff: click-through rate, sessions driven by the notification, maybe bookings if the app had that.”

View Post

Meta·Data Scientist·Technical Phone Screen

Jul 2026
Meta DS interview that was basically a stats and probability gauntlet dressed up as an ads monetization case. Five questions, all quantitative, and the last two caught me pretty off guard because I hadn't thought about mixture distributions in that context before.
  • You run an ads slot with High-Intent and Low-Intent user segments. Given click and conversion probabilities for each segment, revenue per conversion, and cost per impression, compute expected profit per 1,000 impressions under three targeting strategies: show to everyone, show only to users your classifier predicts as High-Intent (with given precision and recall), and show only when a user's posterior conversion probability exceeds some threshold. Derive the profit-maximizing threshold analytically and give its numeric value.
  • For the strategy where you show ads to everyone, compute the variance of the number of conversions per 1,000 impressions using the law of total variance to decompose across the High-Intent and Low-Intent mixture. Does the mixture increase or decrease overdispersion compared to a single Bernoulli with the population-average conversion rate?
  • Your PM suggests targeting impressions exclusively to High-Intent users. Give two quantitative pros and two quantitative cons of this approach, drawing on the numbers from your earlier analysis, including things like reach impact and profit sensitivity to classifier errors.
  • User sessions arrive as a Poisson process at a fixed rate. Write the PDF and CDF of the inter-arrival time distribution, compute its mean, variance, and the probability that a wait exceeds 10 minutes. Then describe what the sample mean of many draws converges to, what theorem governs its approximate distribution for large samples, and how this connects to why average-time-spent metrics stabilize in production dashboards.
  • If real traffic is a mixture of two user types with different click propensities, is the inter-click-time distribution still exponential? If not, what family does it belong to qualitatively, and what diagnostic plot would you use to detect the mixture?

“This was the centerpiece question and I spent way too long on S2 before realizing I needed to account for what the classifier actually does to the composition of shown impressions.”

View Post

Meta·Software Engineer·Technical Phone Screen

Jul 2026
Meta phone screen for a software engineering role, one coding question the whole time. The problem looked deceptively clean but the follow-up pushed into territory I wasn't fully ready for.
  • Given a string, repeatedly remove any maximal contiguous run of identical characters with length at least 2, concatenating the remaining parts each time, until no such run exists. Return the final string. Design a linear-time, linear-space solution, explain why it's correct, and analyze complexity. Follow-up: generalize to removing runs of length at least k for arbitrary k >= 2.

“I got the stack approach pretty quickly: track each character alongside a count, and whenever a count hits 2 (or k in the follow-up), collapse it.”

View Post

Meta·Data Scientist·Onsite - Product Sense / Strategy

Jul 2026
Meta data scientist interview that was basically a compressed product strategy case. You get 10 minutes to walk a VP through a build-vs-don't-build decision, and the scope is wider than it sounds: market sizing, internal capability gaps, privacy risks, and a full experiment roadmap all in one answer.
  • Your VP gives you 10 minutes to make a go or no-go recommendation on building a new product feature from scratch. Walk through your full reasoning, covering market opportunity, internal readiness, and a pre-launch evidence plan.

“This question is basically three separate interviews stapled together.”

View Post