Interview: 'design an unsupervised fraud detection system'.
hardAnswer
- (1) Feature engineering: transaction amount, velocity, device / IP fingerprints, merchant risk, time-of-day, historical account patterns.
- (2) Stack detectors: (a) Isolation Forest on tabular features, (b) autoencoder on account behavior sequences, (c) graph-based (transaction network) for money-laundering rings.
- (3) Ensemble scores.
- (4) Human-in-the-loop: score → analyst → labeled anomalies → semi-supervised uplift.
- (5) Monitor drift + retrain weekly.
- (6) Guardrails: false-positive rate SLO.
Check yourself — multiple choice
- Random
- Feature-eng + stacked detectors (iForest + AE on sequences + graph-based) + ensemble scores + HITL semi-sup + drift monitoring + FP-rate SLO
- Only Isolation Forest
- Just supervise
Fraud AD: feature-eng + stacked detectors + HITL + drift monitoring + SLOs.
#anomaly-detection#interview#applications
Practise Unsupervised Learning
214 interview questions in this topic.
Related questions
- Interview: production fraud detector — daily volume 10M, current FN too high.
- Interview: model trained in region A must now serve region B — approach?
- Interview: 'design anomaly detection for a factory sensor with 200 signals.'
- Interview: 'you're asked to segment 10M customers — how?'
- Anomaly detection in time series — what changes?
- How is drift detection an unsupervised problem?