Cloud / Infra

Move around a Linux box and manipulate files without hesitation.

Asked Shared-terminal screen: 'find the config file for this service and show me the last 20 lines.'

named in the job descriptions

Find the failing signal in a large/rotating log fast — the single highest-value SDET terminal skill.

Asked 'A test is flaky in CI — grep the logs across this directory for the exception and tell me the most frequent error.'

named in the job descriptions

See what's running, what's stuck, and what's holding a port when a test env misbehaves.

Asked 'Your Selenium/Playwright server won't start — port already in use. Walk me through finding and killing what holds it.'

asked in interviews, not named in any job description

Author a small bash script to glue test setup/teardown/CI steps — explicitly mandatory in some JDs.

Asked 'Write a bash script that runs the suite, and if it fails, greps the log and prints the error count.'

named in the job descriptions

Restart a service under test and read its logs.

Asked 'A dependency service died mid-test — how do you restart it and confirm it's healthy?'

asked in interviews, not named in any job description

Fix the 'permission denied' that blocks a script or artifact.

Asked 'Your CI script fails with Permission denied — what do you check and fix?'

asked in interviews, not named in any job description

Get onto a remote test box and pull logs/artifacts off it.

Asked 'A test failed only on the remote runner — how do you get its logs to your machine?'

named in the job descriptions

Explain the container mental model and the layer/cache system in your own words.

Asked 'What's the difference between an image and a container? Why does layer order matter for build speed?'

named in the job descriptions

Build, run, and get inside a container live to debug a test run.

Asked 'Explain docker run -p -v -e. How do you get a shell inside a running container to debug a failing test?'

named in the job descriptions

Package your OWN Python/Playwright test suite so it runs identically anywhere — the single highest-ROI artifact.

Asked 'Write a Dockerfile that containerizes your test suite' (whiteboard or take-home).

named in the job descriptions

Bring up app-under-test + dependencies + test container together with one command.

Asked 'How would you stand up an app, its DB, and your tests together? How do containers on a compose network reach each other?'

named in the job descriptions

Wire the container into CI as a quality gate — the 2026 QA-DevOps convergence expectation.

Asked 'Show me how your tests run in CI. How do they gate a PR merge?'

named in the job descriptions

Speak to how pipelines inject secrets, cache deps, and parallelize test jobs.

Asked 'How do you inject credentials into a pipeline safely? How do you run tests in parallel across jobs?'

named in the job descriptions

Define each core object in one sentence — literacy is now a firm mid-level expectation, not optional.

Asked 'What's a Pod vs a Deployment vs a Service? Where do ConfigMap and Secret fit?'

named in the job descriptions

Survive inside a cluster someone else built — read why a pod/test failed. Web says hands-on (not just concepts) now matters in the screen.

Asked 'A test pod won't start / your test can't reach a service in the cluster — which kubectl commands do you run and in what order?'

named in the job descriptions

Say what each service is and how a test touches it — 'fluent user', not engineer.

Asked 'Which AWS services do your tests interact with, and how?' Expected: S3 artifacts, CloudWatch logs, EC2 runners, Lambda invoke/assert, IAM for access errors.

named in the job descriptions

Use the CLI to move artifacts and configure access for tests.

Asked 'How would you push your test artifacts to S3 from a CI job?'

named in the job descriptions

Not freeze when a JD/interviewer names Azure or GCP instead of AWS — concepts transfer.

Asked 'We're on GCP/Azure, not AWS — is that a problem?' -> map your AWS-for-testing experience onto their equivalents.

named in the job descriptions

Confirm a test triggered the expected system behavior by reading cloud observability — core to the 2026 deployment-verification/monitoring shift.

Asked 'Your test calls an API that fires an async job — how do you verify it actually ran?' -> check the CloudWatch log/metric.

named in the job descriptions

Run cross-browser/cross-device tests on a hosted cloud grid instead of local machines.

Asked 'How do you run cross-browser tests at scale without a local device lab?' -> cloud grid + parallel sessions.

named in the job descriptions

Diagnose why a test can't reach its target (connection refused / timeout / DNS).

Asked 'Your test gets connection refused hitting the app in Docker — how do you debug the networking?'

named in the job descriptions

Know where test credentials come from in a cloud pipeline instead of hardcoding.

Asked 'Where do your tests get their API keys/DB passwords in CI?' -> secrets store, never in code.

named in the job descriptions

Spin up real ephemeral dependencies (Postgres/Redis/Kafka) inside test code with auto-teardown — modern, senior-signaling integration testing.

Asked 'How do you integration-test against a real DB without a shared staging env?' -> Testcontainers ephemeral instance per run.

asked in interviews, not named in any job description

Own one end-to-end infra story with a number — hits his infra gap AND his metrics gap in a single resume line.

Asked 'Tell me about a time you improved test infrastructure' -> the containerize+shard+metric story.

named in the job descriptions