7 Integration Engineer Interview Questions (and How to Answer Them)
The questions that come up in almost every integration engineering interview — with the answers that show senior-level thinking.
Across hundreds of listings and hiring conversations, integration interviews converge on a surprisingly small set of questions. Here are seven that appear constantly, and what a strong answer contains.
1. "The target system is down. What happens to the message?"
The classic. A strong answer covers: retry with exponential backoff, the difference between transient and permanent failures, dead-letter queues, idempotency on redelivery, and alerting. Mention poison messages and you're ahead of most candidates.
2. "When would you choose events over APIs?"
They're testing architectural judgment. Talk about coupling (events decouple producer from consumer), fan-out (one event, many consumers), and the cost (eventual consistency, harder debugging, schema governance). A senior answer includes when not to use events: request/reply semantics, low-volume point-to-point, anything needing an immediate answer.
3. "How do you handle a breaking change in a partner's API?"
Versioning strategy, consumer-driven contracts, anti-corruption layers, and the organisational answer — communication and deprecation windows matter more than technology.
4. "Walk me through securing an API end to end."
OAuth2 client credentials for system-to-system, scopes and claims, mTLS where warranted, secrets management (never in code or logs), rate limiting, and payload-level concerns (PII masking, encryption at rest). Bonus points for mentioning FAPI if it's a banking role.
5. "A batch job that normally takes 20 minutes has been running for 6 hours. Walk me through your investigation."
They want a method, not a guess: check for locks/blocking queries, source data volume anomalies, downstream throttling, infrastructure changes, then the shameful classic — someone changed the query plan. Narrate your telemetry: where would you look first and why.
6. "How do you test integrations?"
Contract tests against API specs, mocked endpoints for unit-level flows, a shared non-prod environment strategy (and its pain), synthetic transactions in production, and reconciliation reports for data integrity. Be honest about the gap between theory and practice here — interviewers know.
7. "Tell me about an integration that failed in production."
A behavioural question wearing a technical costume. Pick a real incident, own your part, and structure it: impact, detection, diagnosis, fix, and — the part that gets you hired — what changed afterwards (monitoring, runbooks, patterns).
Practice these against real role descriptions — our AI Cover Letter Generator and the listings on the board are a good source of the exact language employers use.