ROLE
AI engineering
Compare this role’s stage expectations →
These are the same questions found in the library, grouped here by the capability each one mainly examines.
1175 questions
Questions 101–150 of 1175 on this page.
A · Professional problem solving and implementation14
- How would you design a real-time risk control system? How do rules, features, models, and manual review work together, and how do you trade false blocks against misses?
- How would you design a search system? How do indexing, retrieval, and ranking divide the work, how are Chinese segmentation and features built, and how do you judge how good the search is?
- Tell me about a project you worked on. Why did you choose that technical approach over the others, how was the decision made, and what were its pros and cons?
- How should coupons, discounts, and flash sales integrate with a trading system without slowing the core transaction flow? How would you calculate, combine, and reconcile discounts?
- Public benchmarks keep improving while internal task performance stalls. How would you investigate evaluation distortion?
- The product wants confident AI conclusions, but evidence supports only probabilistic judgments. How would you design communication and traceability?
- After AI output causes actual user harm, product, model, operations, and review teams all take part in the response. How would you divide incident handling and follow-up improvement?
- After replacing a foundation model, overall metrics remain similar but some users report worse results. How would you identify what changed in model behavior and decide whether to roll back?
- You need to sample production AI conversations for emerging risks, but they contain personal or commercial data. How would you design the review?
- A third-party model provider changes its training, retention, or cross-border terms. When should calls pause or receive renewed approval?
- Average AI quality meets the target, but a few high-risk errors can cause harm. How would you set the release gate?
- An AI assistant can call internal tools, but tasks require different permissions. How would you design authorization and audit?
- When the model cannot establish an answer, should the product refuse, return a low-confidence result, or escalate to a person? What distinguishes the choices?
- Overall AI accuracy passes, but one language or user segment has a higher error rate. How would you assess and respond?
No questions in this section match your target. ·
B · Business understanding and domain modeling35
- Renewal fell by six percentage points this quarter, while the share of small customers among new accounts rose. How would you segment customers, usage periods, and service cost to determine whether product value fell or the customer mix changed?
- Besides your product, what else could users do to get the same job done? What does each alternative compete for, and when is it actually the better choice?
- Sales reports growth using signed contract value, but customers will not use the product for months. How would you distinguish contracting, fulfillment, revenue recognition, and customer value, and explain what the team should observe?
- A premium service tier earns more revenue but consumes more human support, and renewal has not improved. How would you determine whether it creates incremental value?
- A product is sold once but requires long-term support and upgrades. Revenue is growing while the service queue accumulates. How would you decide whether these sales still create value?
- Pick a module you own. Explain who it serves, what input it takes, what output it provides, and follow the user task through to the business result.
- When you explain a business, how do you first lay out the whole and then the position and value of the part you own within it? How does someone without a whole-picture view tell it, and what is the difference?
- A customer cancels before service begins, but the contract, product page, and support system describe the cancellation fee differently. How would you determine the refund and make future orders follow a consistent rule?
- An order may use a membership discount, channel offer, and coupon, with exceptions by product, contract, or region. How would you define eligibility and calculation order so customers see a consistent amount and finance can reconcile refunds?
- The business says data must be real time, but no one can explain the cost of a fifteen-minute delay. How would you decide the architecture and freshness target?
- Product, finance, and growth teams each report “active users,” but the numbers differ. What would you ask before deciding whether to standardize the definition?
- A user consented to product analytics, then withdrew consent. Derived metrics already reached reports, experiments, and model-training data. How would you model consent state, purpose scope, and propagation?
- A producer plans to rename a field and change an enum’s meaning, while a dozen teams depend on the current schema. How would you manage the change?
- A user asks to have personal data deleted, while finance must retain transaction records for audit. How would you define the boundary among deletion, anonymization, and traceability?
- In a behavioral event table, what exactly does one row represent? Explain how you settle its grain, and which statistics go wrong when the grain is unclear.
- How does your own business model interact with the industry's or a partner's common model? When fields, definitions, or concepts do not line up, how do you handle it?
- Mobile events may arrive days late, while reports must reflect event time. How would you model and process late, corrected, and duplicate events?
- A critical dataset is reused by several products and reports, but no one can explain its sources, owners, or sensitive fields. How would you improve governance without blocking current use?
- A company buys 100 seats used across subsidiaries. Headquarters signs the contract, but some subsidiaries want separate invoices and permission administrators. How would you distinguish customer, contract, organization, account, member, and seat so access and billing stay consistent?
- Customer master data has separate identifiers and states in CRM, billing, and product systems, so downstream reports disagree. How would you establish explainable sources of truth?
- A quality check occasionally produces false alarms and blocks downstream jobs; removing it once let bad data reach settlement. How would you design tiered handling?
- An account may hold a base plan, temporary trial, campaign benefit, and support credit. Systems implement precedence differently, so access sometimes flips. How would you define eligibility and conflict rules?
- A model feature will be retired, but several online models and historical reports still depend on it. How would you plan retirement and replacement?
- An online feature must refresh every second, but it affects few requests and its maintenance cost keeps rising. How would you decide whether to keep it real time?
- A risk model uses later default as its training label, but many transactions have no final outcome before the observation window ends. How would you handle label maturity, sample inclusion, and retrospective evaluation?
- You are building a 'days active in the last 7 days' feature for a model that predicts next-day churn. How do you define its time semantics so training and online prediction mean the same thing?
- An order has a single status field that means both fulfillment progress and payment result, and several teams read it. How would you split it into two concepts and move old data and old consumers over smoothly?
- A service changes eligibility from “current account status” to “account status at application time plus a risk-review result.” Support needs to explain historical denials, and audit must reproduce the original decision. How would you evolve rules and data models without conflating old and new outcomes?
- Duplicate registration created two customer entities later confirmed to be the same company. Reports should unify them, while contracts, approvals, permissions, and personal actions must retain original responsibility. How would you model the merge?
- A support ticket can be reopened when a customer provides new evidence and may be split into follow-up work. How would you model states and events so closure, reopening, and splitting remain traceable?
- A business rule is corrected retroactively. Reports need corrected results, while decision-makers also need the results available at the time. How would you manage both historical views?
- A producer adds an order status, and an old consumer rejects the whole message because enum validation is strict. Teams disagree over who must update, and consumers have different owners. How would you design compatibility and rollout order?
- Map the area you have owned over time: what users are trying to do, where the main process obstacles are, what your module can influence, and what is outside its control.
- Write a one-page product overview for someone who just joined. How would you mark which statements are fact, which are inference, and which are simply unanswered questions?
- Where does your project sit in the wider business? Who gives it input, who uses its results, and who is affected if it is late or wrong?
No questions in this section match your target. ·
C · Value judgment and direction setting1
No questions in this section match your target. ·