Original practice scenario: A simulated bank receives deposit notifications, ledger postings and reversals on different days. Explain how to reconcile outstanding deposits at a cutoff.
Read model answer +
Reconciliation compares two independently defined event sets at the same cutoff. Joining by customer and amount alone can incorrectly match unrelated transactions or hide delayed postings.
Work through it
Use stable deposit references and accepted event revisions, retaining notification time, posting time and effective date separately.
Aggregate each side to one row per reference before joining; classify matched, pending, reversed and conflicting amounts under explicit rules.
Test duplicate notifications, same-amount deposits, partial adjustments and late postings, and reconcile opening outstanding balances plus movements to closing balances.
Tradeoffs to discuss
A longer matching window reduces false exceptions but delays discovery; unresolved cases should remain visible rather than being forced into matches.
Common pitfalls
Do not exclude unmatched records through an inner join.
Keep cutoff timezone and the treatment of later corrections consistent across both sides.
Original practice scenario: A transaction simulator flags suspicious activity, but confirmed outcomes arrive weeks later and investigators review mostly high-scoring cases.
Read model answer +
Delayed and selective labels distort apparent precision and recall. A credible evaluation must describe which outcomes are observable and avoid treating unreviewed transactions as verified negatives.
Work through it
Freeze features at decision time and record model scores, review selection and eventual outcome timestamps in a versioned evaluation dataset.
Use sufficiently matured cohorts, assess reviewed and representative audit samples separately, and report uncertainty and review-capacity constraints.
Monitor drift in score distributions and segment outcomes, testing threshold policies against operational cost and appeal or correction pathways.
Tradeoffs to discuss
Random audit sampling improves coverage but consumes review resources; importance weighting depends on known nonzero selection probabilities.
Common pitfalls
Do not leak investigator conclusions into training features.
Avoid claiming unbiased recall when the unreviewed population's true outcomes are unknown.
