AI & Technical question
You need to refresh or update or migrate database without affecting clients. How yould you go about it?
- Microsoft
- Capital One
- AI & Technical
- Hard
Practice this question out loud. An AI interviewer asks it, follows up like a real interviewer would, and scores your answer. Type or speak.
Start a mock interview on this question · Mock interview from a job description
What this question tests
Tests technical execution planning for a zero-downtime migration: can you sequence a risky infrastructure change safely as a PM working with engineering.
How to approach it
- Clarify the scope: what kind of migration (schema change, engine upgrade, or moving to a new database entirely), since the risk profile and rollback plan differ significantly.
- Define the non-negotiable constraint: clients must see no downtime and no data loss, so the plan must avoid a hard cutover.
- Propose a phased approach: dual-write to both old and new databases first, backfill historical data, validate consistency between the two, then switch reads to the new database behind a feature flag.
- Plan for rollback: keep the old database live and in sync until the new one is proven stable over a defined validation window, so you can revert instantly if issues appear.
- Coordinate with engineering on monitoring: set up data-consistency checks and error-rate alerts specifically for the migration window, not just general uptime monitoring.
- Communicate with affected internal teams or clients about the migration timeline even though it is designed to be invisible, in case of unexpected issues.
What a strong answer includes
- Uses the standard safe-migration pattern (dual-write, backfill, validate, cutover behind a flag) rather than a risky single-step switch.
- Explicitly plans a rollback path and a validation window before considering the migration complete, showing risk awareness beyond just 'make the change'.
- Separates the write-path migration from the read-path migration, since switching both at once is what usually causes incidents.
- Notes this is a PM's role to sequence priorities and derisk the plan with engineering, not to write the migration code itself.
Common mistakes
- Proposing a single hard cutover, which risks downtime or data loss if anything goes wrong.
- No mention of a rollback plan or validation period before fully committing to the new system.
- Treating this as a purely technical problem with no mention of the PM's role in sequencing and risk communication.
Likely follow-up questions
- How would you decide how long to run dual-write before cutting over fully?
- What would you do if you discovered a data inconsistency mid-migration?
- How would you communicate this migration to stakeholders who are not technical?
More ai & technical questions
- Create an API design for third-party integration for payments.Microsoft · AI & Technical · Hard
- How will you design a web service that enables other developers to create online chat applications?Microsoft · AI & Technical · Hard
- How will you design a mobile version of MS Word and ensure real time collaboration within this app?Microsoft · AI & Technical · Hard
- Write an algorithm for a robot that has to get from point A to point B in a room.Microsoft · AI & Technical · Medium
- Design the security for a chat/messaging application.Microsoft · AI & Technical · Hard
- Design a system that can detect fraudulent use of Microsoft Word.Microsoft · AI & Technical · Hard
More questions from these companies
Learn the skill behind it
Chapters of the AI PM course that teach what this question tests.
- Chapter 1: Foundations: the model and the decisions it forces on you
- Chapter 8: Evals: define good and make the number defensible
- Chapter 6: Agents and agentic architecture