← Meta Interview Insights

Meta·Data Scientist·Onsite - Multi Round·Senior

SeniorPrefer not to say
May 2026

Summary

Brutal Meta DS onsite focused entirely on a News Feed experiment design scenario. Seven sub-questions, all technical, all requiring you to actually know the math. Not a vibe check at all.

Questions Asked (7)

Q1

How would you choose the experimental unit and randomization scheme to reduce interference in a News Feed ranking experiment, given a 10% treatment cap and a highly skewed degree distribution?

A/B Testing & ExperimentationTechnical Trade-offs
Author's notes

This is where I spent most of my mental energy.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by acknowledging the interference problem in social networks and the constraints (10% cap, skewed degrees). Then propose a randomization scheme that balances interference control with practical feasibility, such as cluster randomization on the graph or ego-network randomization, and discuss trade-offs like bias, variance, and power.

Pro tip: Emphasize that the choice of experimental unit should align with the interference mechanism: if interference is mainly through close connections, randomize at the cluster level (e.g., communities) to contain spillover; if it's global, consider a switchback or time-based randomization. Also, mention that with a 10% cap, you might need to use a biased but low-variance design like graph cluster randomization with a small number of clusters, and correct for bias via post-stratification.

1. Define interference and constraints

Clarify the interference mechanism (e.g., spillover through social ties) and the constraints: 10% treatment cap and skewed degree distribution. Discuss how these affect the choice of experimental unit.

2. Evaluate candidate randomization units

Consider units like users, ego-networks, communities, or time-based switchbacks. For each, assess how well they contain interference, their feasibility under the 10% cap, and their statistical properties (bias, variance, power).

3. Choose a scheme and justify trade-offs

Select a scheme (e.g., cluster randomization on communities) and explain why it balances interference reduction with the cap and skew. Discuss how to handle the skewed degree distribution, e.g., by weighting or stratifying by degree.

4. Address implementation and analysis

Outline how to implement the randomization (e.g., graph partitioning algorithms) and analyze results (e.g., cluster-robust standard errors, bias correction). Mention potential pitfalls like dilution and network effects.

Key Points to Mention

  • Interference/spillover effects in social networks
  • Cluster randomization (e.g., graph cluster randomization, ego-network randomization)
  • Trade-offs between bias and variance in design
  • Handling skewed degree distribution (e.g., stratification, weighting)
  • 10% treatment cap implications (e.g., limited power, need for efficient design)
  • Alternative designs like switchback or time-based randomization
  • Analysis considerations: cluster-robust SEs, bias correction, dilution

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q2

Define the direct effect on consumers, the spillover effect, and the total effect as precise estimands. What exposure model makes these identifiable?

A/B Testing & ExperimentationProduct Analytics & Metrics
Author's notes

I knew what spillover meant conceptually but defining it as a formal estimand with exposure conditions was harder than I expected.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by defining the causal estimands in a potential outcomes framework, clearly distinguishing direct, spillover, and total effects. Then introduce an exposure model (e.g., partial interference with a known exposure mapping) that makes these effects identifiable, and discuss assumptions and estimation strategies. Finally, connect to practical A/B testing scenarios at Meta, such as network effects in social experiments.

Pro tip: Emphasize that spillover effects are only identifiable under specific exposure models, and that ignoring them can lead to biased estimates of the total effect. Mention that in practice, cluster randomization or graph-based exposure models are often used to mitigate interference.

1. Define potential outcomes and causal estimands

Formally define the direct effect as the difference in outcomes for an individual under treatment vs. control, holding the treatment assignment of others fixed. Define spillover effect as the difference in outcomes for an individual due to changes in the treatment assignment of others, holding their own treatment fixed. Define total effect as the difference between the outcome when an individual is treated and others are treated in some way vs. when the individual is untreated and others are untreated.

2. Introduce exposure model and identifiability assumptions

Describe an exposure model that maps the treatment assignment vector to an individual's exposure. For example, partial interference with a known exposure mapping (e.g., the proportion of treated neighbors). State assumptions like consistency, no unmeasured confounding, and correct specification of the exposure model that make the estimands identifiable from observed data.

3. Discuss estimation and inference

Explain how to estimate these effects using methods such as inverse probability weighting, outcome regression, or machine learning under the exposure model. Mention challenges like interference and the need for cluster-robust standard errors.

4. Connect to practical experimentation at Meta

Relate the concepts to real-world A/B testing at Meta, where interference is common due to social networks. Discuss strategies like cluster randomization, graph cluster randomization, or using exposure models to adjust for spillovers in experiments.

Key Points to Mention

  • Potential outcomes framework (Rubin causal model) and SUTVA violation
  • Direct effect: effect of own treatment holding others' treatment fixed
  • Spillover effect: effect of others' treatment holding own treatment fixed
  • Total effect: combined effect of own and others' treatment
  • Exposure model: mapping from treatment assignment vector to exposure (e.g., partial interference, network exposure)
  • Identifiability assumptions: consistency, no unmeasured confounding, correct exposure mapping
  • Estimation methods: IPW, outcome regression, machine learning
  • Practical implications: cluster randomization, graph cluster randomization, network A/B testing

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q3

Walk through how you'd compute point estimates and 95% confidence intervals using cluster-robust or randomization-inference standard errors. What assumptions are you making and what breaks if they're violated?

A/B Testing & Experimentation
Author's notes

Felt more comfortable here.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clarifying the experimental design (e.g., cluster randomization, user-level randomization with clustered data) and the estimand (e.g., average treatment effect). Then describe the point estimate (difference in means or weighted average) and how to compute standard errors using cluster-robust or randomization inference, explicitly stating assumptions and consequences of violations.

Pro tip: Emphasize that cluster-robust SEs are asymptotic and rely on a sufficient number of clusters; with few clusters, consider bootstrap or randomization inference. Also, mention that randomization inference is exact under the sharp null and can be more robust when clusters are few or unbalanced.

1. Clarify design and estimand

Identify the randomization unit (e.g., user, cluster) and the outcome metric. Define the causal estimand (e.g., ATE) and the point estimate (e.g., difference in means or regression coefficient).

2. Compute point estimate

Calculate the point estimate as the difference in average outcomes between treatment and control, or via regression with treatment indicator. If clusters are unequal, consider weighting by cluster size.

3. Choose SE method: cluster-robust or randomization inference

For cluster-robust SEs, use the sandwich estimator with cluster-level sums of scores. For randomization inference, permute treatment assignment at the cluster level and compute the distribution of the test statistic under the sharp null.

4. Construct 95% confidence interval

For cluster-robust, use normal approximation: estimate ± 1.96 * SE. For randomization inference, use the 2.5th and 97.5th percentiles of the permutation distribution of the estimator (or invert the test).

5. State assumptions and violations

List key assumptions: clusters are independent, large number of clusters for asymptotic validity, correct model specification for cluster-robust; for randomization inference, sharp null and exchangeability. Discuss what breaks: few clusters lead to undercoverage; violations of independence (e.g., interference) bias SEs; model misspecification affects cluster-robust.

Key Points to Mention

  • Cluster-robust standard errors account for within-cluster correlation but assume clusters are independent and the number of clusters is large enough for asymptotic normality.
  • Randomization inference provides exact p-values and confidence intervals under the sharp null hypothesis, without relying on large-sample approximations, but requires the ability to permute treatment assignment.
  • With few clusters, cluster-robust SEs can be downward biased; use small-sample corrections (e.g., CR2) or alternative methods like bootstrap or randomization inference.
  • Point estimates should be weighted appropriately if clusters are of unequal size, e.g., by cluster size, to estimate the average treatment effect at the user level.
  • Violations of independence (e.g., interference between clusters) can invalidate both methods; consider design-based solutions or sensitivity analysis.
  • For cluster-robust SEs, the choice of clustering level matters: cluster at the randomization unit to avoid over-rejection.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q4

Derive the required sample size under clustering with an average cluster size of 120 and ICC of 0.02. Use the design effect formula to find the effective sample size needed for 80% power to detect a 0.6% relative lift, given a baseline mean of 4.0 sessions per user and SD of 5.5.

A/B Testing & Experimentation
Author's notes

Okay so I actually like this kind of question because it's just math.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

First, compute the required sample size for an individually randomized design using the baseline mean, standard deviation, and target relative lift to achieve 80% power. Then, adjust for clustering by calculating the design effect (1 + (average cluster size - 1) * ICC) and multiply the individual sample size by this factor to get the required number of users. Finally, divide by the average cluster size to determine the number of clusters needed.

Pro tip: Always clarify whether the sample size refers to users or clusters, and mention that the design effect inflates the variance, so you need more clusters to maintain power. Also, note that the ICC is typically small in online experiments, but even a small ICC can substantially increase sample size when clusters are large.

1. Compute individual sample size

Calculate the required sample size per arm assuming no clustering, using the formula for comparing means: n = 2 * (Z_alpha/2 + Z_beta)^2 * SD^2 / (delta)^2, where delta is the absolute lift (baseline mean * relative lift).

2. Calculate design effect

Use the design effect formula: DE = 1 + (m - 1) * ICC, where m is the average cluster size (120) and ICC is 0.02. This quantifies the variance inflation due to clustering.

3. Adjust sample size for clustering

Multiply the individual sample size by the design effect to get the required number of users (or observations) under clustering. This is the effective sample size needed.

4. Convert to number of clusters

Divide the adjusted sample size by the average cluster size (120) to determine the number of clusters required per arm. Round up to the nearest whole number.

5. Sanity check and interpret

Verify that the numbers make sense (e.g., design effect >1, sample size larger than individual case) and discuss practical implications such as feasibility and potential for unequal cluster sizes.

Key Points to Mention

  • Design effect formula: DE = 1 + (m - 1) * ICC
  • Sample size formula for two-sample t-test: n = 2 * (Z_{1-α/2} + Z_{1-β})^2 * σ^2 / Δ^2
  • Absolute lift calculation: Δ = baseline mean * relative lift = 4.0 * 0.006 = 0.024
  • Standard values: Z_{0.975} ≈ 1.96, Z_{0.80} ≈ 0.84, so (1.96+0.84)^2 ≈ 7.84
  • Clustering increases required sample size; effective sample size = individual sample size * DE
  • Number of clusters = total required users / average cluster size, rounded up

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q5

Propose at least three concrete diagnostic checks to detect leakage or interference in the experiment, and explain how you'd respond to each finding.

A/B Testing & ExperimentationRoot Cause Analysis
Author's notes

Cut-edge exposure rate was my first one, treated-producer content share seen by control users was second.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by framing the goal: to ensure experiment validity by detecting leakage or interference. Then, for each diagnostic check, describe the method, the expected outcome under no leakage/interference, and the action plan if the check fails. Emphasize a systematic, data-driven approach to root cause analysis and mitigation.

Pro tip: Prioritize checks that are easy to implement and interpret, such as Sample Ratio Mismatch (SRM) and pre-experiment covariate balance, as they often reveal subtle issues. Also, consider the trade-offs between sensitivity and specificity of each check.

1. Sample Ratio Mismatch (SRM) Check

Compare the observed ratio of users in treatment and control groups to the expected ratio using a chi-squared test. A significant deviation indicates potential leakage or assignment issues.

2. Pre-experiment Covariate Balance Check

Compare key pre-experiment metrics (e.g., age, tenure, past behavior) between groups using statistical tests or effect sizes. Imbalance suggests interference or non-random assignment.

3. Treatment Effect on Guardrail Metrics

Analyze metrics that should not be affected by the treatment (e.g., user demographics, pre-period behavior). Significant differences indicate leakage or interference.

4. Response to Findings

For each failed check, investigate root causes (e.g., logging errors, cross-contamination, network effects). Mitigate by fixing bugs, re-randomizing, or using techniques like cluster randomization or isolation.

Key Points to Mention

  • Sample Ratio Mismatch (SRM) and its implications
  • Pre-experiment covariate balance and statistical tests (e.g., t-test, chi-squared)
  • Guardrail metrics and their role in detecting interference
  • Root cause analysis techniques (e.g., segment analysis, logging audits)
  • Mitigation strategies: re-randomization, cluster randomization, isolation, or excluding contaminated users
  • Importance of documenting and communicating findings to stakeholders

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q6

Which sequential testing method would you use for interim looks every two days, and what are your stopping and continuation rules to maintain Type I error control?

A/B Testing & ExperimentationTechnical Trade-offs
Author's notes

Went with alpha-spending using an O'Brien-Fleming boundary because it's conservative early and I've actually used it before.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by acknowledging the need for a sequential testing method that allows for frequent interim looks while controlling Type I error. Recommend the alpha spending approach (e.g., O'Brien-Fleming or Pocock) or group sequential methods, and specify how you would set stopping boundaries and continuation rules. Emphasize the trade-offs between early stopping and error control, and mention practical considerations for implementation at Meta's scale.

Pro tip: Mention that with looks every two days, you need to account for the number of looks and the timing, and that alpha spending functions like O'Brien-Fleming are often preferred for their conservative early stopping, which is useful when metrics need time to stabilize. Also, note that you would pre-register the analysis plan to avoid p-hacking.

1. Choose the sequential testing framework

Select a group sequential design with alpha spending, such as O'Brien-Fleming or Pocock boundaries, or a continuous monitoring approach like always valid inference. Justify based on the need for frequent looks and Type I error control.

2. Define the alpha spending function

Specify how alpha is allocated across interim looks. For example, O'Brien-Fleming spends little alpha early and more later, while Pocock spends equally. This determines the stopping boundaries.

3. Set stopping and continuation rules

Establish boundaries for efficacy and futility at each look. If the test statistic crosses the efficacy boundary, stop for success; if it crosses futility, stop for no effect; otherwise continue to the next look.

4. Compute boundaries and adjust p-values

Use software (e.g., gsDesign in R) to compute exact boundaries based on the spending function, number of looks, and information fractions. Adjust p-values and confidence intervals for sequential monitoring.

5. Address practical considerations

Discuss handling of uneven look intervals, missing data, and multiple metrics. Mention that you would pre-register the plan and monitor for any deviations.

Key Points to Mention

  • Type I error control: alpha spending functions maintain overall Type I error at the desired level despite multiple looks.
  • O'Brien-Fleming vs. Pocock boundaries: trade-offs between early stopping and power.
  • Information fraction: the proportion of total planned information at each interim look.
  • Stopping rules: efficacy and futility boundaries, and how they affect decision-making.
  • Software tools: gsDesign, rpact, or equivalent for boundary computation.
  • Pre-registration: importance of pre-specifying the analysis plan to avoid bias.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q7

If leadership forces a standard 50/50 user-level randomization instead of clustering, what post-hoc approaches would you use to bound the bias and produce conservative confidence intervals?

A/B Testing & ExperimentationAdaptability & Ambiguity
Author's notes

This one felt like a trick question at first because the setup is basically 'what do you do when you can't fix the design.' I mentioned CUPED with pre-period to reduce variance, exposure-weighted IV as a way to isolate the complier effect, and sensitivity bounds to show a range of plausible ATEs under different interference assumptions.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Acknowledge the constraint and focus on post-hoc methods to quantify and bound bias from ignoring clustering. Use cluster-robust variance estimation, sensitivity analysis, and conservative adjustments to produce valid confidence intervals.

Pro tip: Emphasize that while you can't eliminate bias post-hoc, you can bound it and be transparent about assumptions. Show you understand the trade-off between validity and precision.

1. Diagnose clustering

Assess the degree of clustering in the data by computing intra-cluster correlation (ICC) for key metrics. This quantifies how much user-level randomization violates independence.

2. Cluster-robust inference

Apply cluster-robust standard errors (e.g., CR2 or bootstrap) at the cluster level to account for within-cluster correlation. This yields conservative confidence intervals.

3. Sensitivity analysis

Perform sensitivity analysis by varying assumptions about cluster effects (e.g., using different ICC values) to see how conclusions change. Report the range of treatment effects.

4. Conservative adjustments

Use methods like the delta method or variance inflation factors to adjust point estimates and intervals. Consider worst-case bounds by assuming maximum plausible correlation.

5. Communicate uncertainty

Clearly document the limitations and present results with conservative intervals. Recommend follow-up experiments with proper randomization if possible.

Key Points to Mention

  • Intra-cluster correlation (ICC) and design effect
  • Cluster-robust standard errors (e.g., CR2, bootstrap)
  • Sensitivity analysis for unmeasured clustering
  • Variance inflation factor and effective sample size
  • Trade-off between bias and precision
  • Transparency in reporting and decision-making

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.