ROLE
Data engineering
Compare expectations across career stages →
These are the same questions found in the library, grouped here by the capability each one mainly examines.
453 questions
Extended questions 301–350 of 453 on this page.
A · Professional problem solving and implementation5
- A model begins returning structured JSON, while mobile, server, and analytics consumers upgrade at different rates; new fields may be absent or enum values may change. How would you define the output contract, handle validation failures, and preserve compatibility?
- An AI assistant reads user-uploaded documents that may tell it to ignore existing rules and call external tools. How would you distinguish data from instructions, constrain tool permissions, and verify the safeguards?
- Overall task success is unchanged after a model update, but outcomes diverge more across users by language, device, and experience. How would you define evaluation segments, manage small-sample risk, and decide whether to continue rollout?
- Users can upload archives for processing, and a past incident showed that malicious files could exhaust resources or access paths without authorization. How would you design isolation, extraction limits, content checks, and safe failure behavior?
- The team needs real user conversations to reproduce model failures, but logs contain personal and sensitive business content. How would you minimize collection, redact data, authorize access, and set retention while preserving debugging value?
No questions in this section match your target. ·
B · Business understanding and domain modeling40
- A user upgrades mid-cycle and later requests a refund. The contract requires a calculation based on remaining service time, while promotional discounts and taxes also affect settlement. How would you decompose the rules and make the amount calculation traceable?
- A campaign allows each user to claim a discount once, but under high concurrency a user submits from multiple devices and receives several entitlements. How would you map the eligibility rule to atomic reservation, issuance records, and failure compensation?
- A user clicks a campaign link on Monday and purchases on Tuesday, but the purchase event uploads on Friday because the client was offline. How would you define event time, arrival time, attribution windows, and backfill rules so results remain explainable?
- A database field has been replaced, but reports, exports, and external integrations still read it; deleting it directly could cause silent errors. How would you plan compatibility migration, consumer confirmation, and final removal?
- A product feature is configured by customer contract, adjusted by experiment group, and constrained by regional rules; analysts need to explain which combination shaped a user's experience. How would you model configuration sources, scope, effective time, and decision provenance?
- Customers pay by API usage, but usage events may be duplicated, arrive late, or be replayed at month-end; sent invoices can lead to disputes. How would you define billable units, deduplication keys, close windows, and corrections?
- A dashboard's total conversion rate cannot explain which customer segment or process step caused a change; teams want to drill down by channel, plan, and region, but dimensions have different grains. How would you define fact grain and valid aggregation rules?
- A promotion increases subscriptions, but many subscribers renew at a lower price and full-price renewal has not yet reached its observation window. How would you decompose acquisition, renewal, discount, and churn to judge long-term value?
- Order revenue is shared by a platform, provider, and channel partner. Weeks later, part of the order is refunded; one party has already settled and lacks sufficient balance. How would you express allocation rules, refund responsibility, recovery, and reconciliation state?
- Business reporting needs the duration of every state transition, while an operations page needs only the object's current state; a wide table overwrote history during updates. How would you separate event facts, snapshots, and current-state models?
- CRM, billing, and product logs use customer IDs, contract IDs, and anonymous device identifiers; analysts need one customer's lifecycle. How would you model identity mappings, confidence, temporal validity, and correction?
- Several teams maintain customer profiles. When data is wrong, no one knows which system is authoritative, and systems also write values back to one another. How would you define field ownership, source of truth, synchronization direction, and dispute handling?
- A device reports a cumulative counter, but the counter resets after device replacement or data correction; the data platform may also backfill historical events. How would you define deltas, reset markers, negative differences, and recomputation boundaries?
- A numeric feature value of zero may mean the user had no activity, missing may mean the device did not report, and unknown may mean computation is unfinished; the pipeline currently fills all three with zero. How would you define the semantics and downstream handling?
- A global product counts daily activity by calendar day, but product analytics, the warehouse, and model serving use user time, UTC, and server time respectively; results often differ near month-end. How would you define boundaries and versioned semantics?
- Billing changes from calendar-month pricing to service-day pricing, with the new rule applying only to new contracts; old contracts cross the release date, and support still needs to explain past invoices. How would you version rules, contract applicability, and invoice snapshots?
- Business categories may be added, merged, or renamed. Reports must reproduce the definition used at the time and compare results under the current definition. How would you preserve source categories, mapping versions, and report definitions?
- An event producer wants to change a field from a string to a structured object, but some consumers deploy slowly and others drop unknown fields when writing records back. How would you stage dual writes, version negotiation, migration, and retirement of the old format?
- A customer-ownership field in historical events is wrong and must be corrected in bulk to recompute downstream metrics; the same events may also trigger notifications and partner callbacks. How would you separate data recomputation from side-effect replay?
- Active users and revenue are growing, but human support hours, refunds, and infrastructure costs per user are rising faster. How would you use user segments, tasks, and service flows to assess the economic quality of growth?
- An enterprise pays per active seat, but some employees share accounts across shifts while others have accounts they never use; the contract does not define “active.” How would you determine the billable entity, activity definition, and dispute evidence?
- Orders contain nested line items, shipping, and tax, while refunds may apply to only some items; joining order headers directly to lines doubles totals in existing reports. How would you model grain and aggregation semantics?
- Sales performance rolls up through a regional hierarchy. Midyear, a region is split; historical reports must remain unchanged, while future reports use the new hierarchy. How would you represent organization versions, effective time, and attribution?
- Some transaction details must be retained for years under policy, while product analytics wants to use them for user segmentation and experiments. How would you model purpose authorization, retention, and access views?
- A user's transaction enters a behavioral feature window, then is reversed days later due to refund or fraud review; downstream models and reports have already consumed the feature. How would you model reversal, window recomputation, and impact propagation?
- A team wants a ratio feature of completions divided by applications, but applications can be withdrawn, duplicated, or later invalidated beyond the observation window. How would you define numerator, denominator, maturity window, and recomputation semantics?
- New users have little behavioral history, and the feature pipeline fills missing values with a global average, making “not observed” indistinguishable from “average behavior.” How would you represent cold-start status and evidence boundaries?
- An online feature service can return its last computed value after a cache expires, but some predictions need the freshest feature while others tolerate brief staleness. How would you define freshness metadata, task constraints, and unavailable-value fallback?
- A new workflow requires every record to have an ownership field, but historical data lacks it and older clients still write records. How would you design backfill provenance, dual-read/write migration, validation gates, and retirement of old clients?
- An order is priced in one currency and paid in another; a refund occurs after the exchange rate changes, and line-item rounding leaves the total off by the smallest currency unit. How would you define rate timing, rounding order, and ownership of the difference?
- A user requests personal-data deletion, but the system uses append-only events, search indexes, and offline analytics copies; some events must remain for business audit. How would you distinguish deletable fields, identity mappings, cryptographic erasure, and retention exceptions?
- A click-through feature suggests low interest in an item, but it appeared briefly at the bottom of a page; the team still trains a recommender on clicks. How would you model exposure opportunity, position, visibility duration, and user action separately?
- A prediction formerly used individual history, but a product now recommends at the organization level; membership changes over time and a person may belong to several organizations. How would you define feature ownership, membership validity, and aggregation deduplication?
- After an upstream migration, a feature keeps falling even though business activity is stable; consumers did not know the source field had been deprecated. How would you define feature lineage, health state, and replacement-source cutover?
- The system changes its user key from email to an immutable ID, but mobile caches, partner callbacks, and historical exports still reference email. How would you design mappings, a dual-read period, privacy cleanup, and retirement of old references?
- The team wants to move eligibility checks from service code to a configurable rules engine so Operations can adjust them; many historical exceptions exist and execution order is implicit in code. How would you discover rules, verify semantic equivalence, and govern configuration access?
- An early consumer deduplicates only by order ID, but one order can have several legitimate changes; the key must include an event version and historical messages need replay. How would you migrate keys without losing or duplicating processing?
- A customer entity must be split into two real customers after an erroneous merge, while old orders, permissions, and metrics reference the original ID. How would you model the split, historical attribution, current-reference migration, and indeterminate records?
- Several teams define user-status events differently, and the platform plans to unify them; some consumers rely on special meanings in old fields and cannot migrate at once. How would you build compatibility adapters, parallel validation, and deprecation governance?
- A new model requires additional features and changes the meaning of existing ones, while online serving must still support rollback to the old model. How would you version feature contracts, deployment order, dual-read cost, and retirement criteria?
No questions in this section match your target. ·
C · Value judgment and direction setting5
- A new operations dashboard is visited often, but interviews show most people only take screenshots before reviews and do not adjust resources or actions. How would you decide whether it creates decision value or only serves a display need?
- A redesign improves average conversion but increases task failure for a constrained user group; expanding the sample requires more research time. How would you prioritize the risk and decide whether to pause rollout?
- Policy and procurement cycles in a new market are unclear, yet the company must set annual goals and budgets. How would you translate long-term direction into quarterly hypotheses, validation milestones, and triggers for adjusting the plan?
- The team plans to pilot a service in three cities, but supply quality, repeat use, and unit economics are unproven; sales materials already show a firm expansion timeline. How would you rewrite stage goals and external commitments?
- The team defines conversion as purchasers divided by visitors. After a redesign, visits fall while purchaser count stays roughly flat, so the rate rises. How would you decide whether the redesign improved the target user's purchase experience?
No questions in this section match your target. ·