← Meta Interview Insights

Meta·Data Scientist·Technical Phone Screen·Senior

Senior
Jun 2026

Summary

A technical screen for a Data Scientist role at Meta that was basically a deep dive into a single fraud classifier scenario. Four sub-parts, all connected, and they expected you to carry context from one into the next. Not a casual conversation.

Questions Asked (4)

Q1

You inherit a binary fraud classifier. Given a holdout confusion matrix with TP=200, FP=800, FN=100, TN=99,900, compute precision, recall, F1, and false-positive rate.

Product Analytics & MetricsTechnical Trade-offs
Author's notes

The numbers look clean but the class imbalance is jarring when you actually work through them.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by clearly stating the formulas for precision, recall, F1, and false-positive rate, then plug in the given values step by step. Compute each metric carefully, and interpret the results in the context of fraud detection, highlighting the trade-off between precision and recall.

Pro tip: After computing the metrics, discuss the business implications: in fraud detection, a high false-positive rate can erode customer trust and increase operational costs, so it's crucial to balance precision and recall based on the cost of errors.

1. Define the metrics

State the formulas: Precision = TP / (TP + FP), Recall = TP / (TP + FN), F1 = 2 * (Precision * Recall) / (Precision + Recall), False Positive Rate = FP / (FP + TN).

2. Plug in the values

Substitute the given numbers: TP=200, FP=800, FN=100, TN=99,900 into the formulas.

3. Calculate each metric

Compute precision = 200/(200+800)=0.2, recall = 200/(200+100)=0.6667, F1 = 2*(0.2*0.6667)/(0.2+0.6667)=0.3077, FPR = 800/(800+99900)=0.00794.

4. Interpret the results

Explain what these numbers mean: low precision (20%) means many false positives, moderate recall (66.7%) means missing some fraud, F1 is low due to poor precision, and FPR is low (0.79%) because of large TN.

5. Discuss trade-offs and business impact

Relate to fraud detection: high FP may inconvenience customers, missing fraud (FN) may cause financial loss. Suggest possible adjustments like threshold tuning to balance precision and recall.

Key Points to Mention

  • Precision = TP / (TP + FP) = 200/1000 = 0.2
  • Recall = TP / (TP + FN) = 200/300 ≈ 0.667
  • F1 = 2 * (Precision * Recall) / (Precision + Recall) ≈ 0.308
  • False Positive Rate = FP / (FP + TN) = 800/100700 ≈ 0.00794
  • Interpretation: Low precision indicates many false positives; recall is moderate; F1 is low due to poor precision.
  • Business context: In fraud detection, false positives can block legitimate transactions, while false negatives miss fraud. Balance depends on cost of each error.

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

Q2

False negatives cost $20 and false positives cost $0.20, with 1% fraud prevalence. How do you pick a threshold using predicted probabilities to maximize expected utility, and which metric, PR-AUC or ROC-AUC, better reflects improvements at low prevalence and why?

Product Analytics & MetricsTechnical Trade-offsRoot Cause Analysis
Author's notes

This is where it got interesting.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by framing the problem as expected utility maximization: for each predicted probability p, compute the expected cost of predicting positive vs. negative, and choose the threshold that minimizes total expected cost. Then discuss why PR-AUC is more informative than ROC-AUC when prevalence is low, because ROC-AUC can be overly optimistic due to the large number of true negatives.

Pro tip: Mention that the optimal threshold depends on the cost ratio and prevalence, and that in practice you should calibrate probabilities first and consider business constraints like alert volume. Also, note that PR-AUC focuses on the positive class and is more sensitive to changes in false positives when prevalence is low.

1. Define the expected cost function

For a given threshold t, total expected cost = 20 * FN + 0.20 * FP, where FN and FP are counts based on predicted probabilities >= t. Express this in terms of predicted probabilities and true labels.

2. Derive the optimal threshold

For each instance with predicted probability p, predict positive if expected cost of positive (0.20 * (1 - p)) is less than expected cost of negative (20 * p). Solve for p: 0.20*(1-p) < 20*p => p > 0.20/(20+0.20) ≈ 0.0099. So threshold ≈ 0.01.

3. Validate with prevalence and cost ratio

Note that the optimal threshold is cost ratio / (cost ratio + 1) adjusted for prevalence? Actually, the derivation above assumes we compare expected costs per instance, but prevalence affects the base rates. Re-derive using Bayes: predict positive if P(fraud|p) * cost_FN > (1-P(fraud|p)) * cost_FP. With calibrated probabilities, P(fraud|p) = p. So threshold = cost_FP / (cost_FN + cost_FP) = 0.20/(20+0.20) ≈ 0.0099. Prevalence does not directly enter if probabilities are calibrated, but it affects the distribution of p.

4. Compare PR-AUC vs ROC-AUC

Explain that ROC-AUC is invariant to class prevalence and can be misleadingly high when negatives dominate. PR-AUC (precision-recall AUC) focuses on the positive class and is more sensitive to false positives, making it better for imbalanced settings.

5. Conclude with practical implications

Summarize that the threshold should be set to minimize expected cost, and that PR-AUC is preferred for model selection and monitoring when prevalence is low, as it better reflects improvements in identifying the minority class.

Key Points to Mention

  • Expected utility maximization: minimize 20*FN + 0.20*FP
  • Optimal threshold formula: cost_FP / (cost_FN + cost_FP) when probabilities are calibrated
  • Prevalence affects the distribution of predicted probabilities and the baseline performance
  • ROC-AUC is insensitive to class imbalance and can be overly optimistic
  • PR-AUC focuses on the positive class and is more informative for low prevalence
  • Calibration of predicted probabilities is crucial for threshold optimization

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

Q3

How would you check whether the classifier's predicted probabilities are well-calibrated, and how does calibration affect threshold selection?

Technical Trade-offsProduct Analytics & Metrics
Author's notes

Reliability plot first: bin predictions, plot mean predicted probability against actual fraction of positives per bin.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by defining calibration and explaining why it matters for decision-making. Then describe methods to assess calibration, such as reliability diagrams and metrics like Brier score or ECE. Finally, discuss how calibration impacts threshold selection, emphasizing that well-calibrated probabilities allow for optimal thresholds based on costs and benefits.

Pro tip: Mention that calibration should be evaluated on a held-out set and that different calibration methods (Platt scaling, isotonic regression) have trade-offs. Also, note that calibration can vary across subgroups, which is crucial for fairness at Meta.

1. Define calibration and its importance

Explain that calibration measures how well predicted probabilities reflect true likelihoods. Emphasize its role in reliable decision-making and threshold optimization.

2. Methods to check calibration

Describe visual tools like reliability diagrams and quantitative metrics like Expected Calibration Error (ECE) and Brier score. Mention the need for a validation set.

3. Calibration techniques

Briefly discuss post-hoc calibration methods such as Platt scaling and isotonic regression, and their trade-offs (e.g., parametric vs. non-parametric).

4. Impact on threshold selection

Explain that well-calibrated probabilities enable setting thresholds based on expected costs and benefits. Poor calibration can lead to suboptimal thresholds and biased decisions.

5. Practical considerations

Highlight the importance of evaluating calibration on subgroups, monitoring calibration over time, and considering business metrics when selecting thresholds.

Key Points to Mention

  • Reliability diagrams (calibration curves) and how to interpret them
  • Expected Calibration Error (ECE) and Brier score as quantitative metrics
  • Platt scaling and isotonic regression for calibration
  • The relationship between calibration and threshold optimization (e.g., cost-sensitive thresholds)
  • The effect of imbalanced data on calibration and threshold selection
  • Subgroup calibration and fairness implications

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

Q4

Design an online evaluation plan that improves fraud catch rate without over-blocking legitimate users. Include guardrails, a shadow or staged rollout approach, success criteria, and rollback triggers.

A/B Testing & ExperimentationSystem DesignProduct Analytics & Metrics
Author's notes

I went with a shadow mode first: new model scores transactions in parallel but doesn't act, so you can compare distributions without any user impact.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by defining the problem and key metrics (fraud catch rate, false positive rate, precision/recall), then propose a staged evaluation plan with shadow mode, guardrails, and rollback triggers. Emphasize the trade-off between catching fraud and over-blocking legitimate users, and outline success criteria and monitoring.

Pro tip: Frame the plan as a continuous optimization loop: use shadow mode to gather data without impacting users, then gradually roll out with strict guardrails and automated rollback to balance risk and learning.

1. Define Objectives and Metrics

Clarify the primary goal: improve fraud catch rate while keeping false positive rate (over-blocking) within acceptable bounds. Define metrics like precision, recall, F1, and business impact (e.g., user friction).

2. Design Shadow and Staged Rollout

Start with shadow mode: run the new model in parallel without affecting decisions, compare against current system. Then stage rollout: small percentage of traffic, gradually increase while monitoring.

3. Establish Guardrails and Success Criteria

Set guardrail metrics (e.g., false positive rate, user complaints, latency) that must not degrade. Define success criteria: e.g., 10% relative improvement in fraud catch rate with no more than 0.1% increase in false positives.

4. Define Rollback Triggers and Monitoring

Specify automatic rollback triggers: if guardrails breached (e.g., false positive rate exceeds threshold, latency spikes), revert to previous model. Set up real-time monitoring and alerting.

5. Analyze and Iterate

After staged rollout, analyze results, gather feedback, and iterate on model or thresholds. Document learnings and plan next steps.

Key Points to Mention

  • Trade-off between fraud catch rate and false positives (precision-recall trade-off)
  • Shadow mode for safe evaluation without user impact
  • Staged rollout with gradual traffic increase (e.g., 1%, 5%, 20%, 50%, 100%)
  • Guardrail metrics: false positive rate, user complaints, latency, revenue impact
  • Success criteria: statistically significant improvement in fraud catch rate with acceptable false positive increase
  • Rollback triggers: automated alerts when guardrails breached, with clear thresholds

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