Infrastructure Engineer interview questions and how to answer them
Infrastructure interviews test whether you can keep things running when they're on fire, not whether you've memorized flags. Expect a mix of systems questions, a live troubleshooting exercise and a lot of talk about outages you've lived through. Here's what gets asked and what the interviewer is listening for.
What happens in each round
- 1
Recruiter screen
What happens
Whether your stack matches theirs. They'll ask which cloud you've used, whether you've written Terraform or Ansible, and how you feel about on-call. Keep answers short and use their tool names.
- 2
Hiring manager call
What happens
The scope of what you've owned. They want a system you ran end to end, how big the blast radius was when it broke, and whether you talk about the team or only yourself.
- 3
Live troubleshooting or Linux exercise
What happens
How you debug with a shell open. You might get a broken container, a server that can't reach a database, or a slow web app. They care about your order of checks and whether you say your reasoning out loud.
- 4
Infrastructure design round
What happens
Whether you can sketch a working environment on a whiteboard: networks, load balancers, databases, secrets, monitoring and a way to deploy. They push on failure cases and cost.
- 5
Behavioral and incident review with the team
What happens
How you act during and after an outage. Expect questions about a postmortem you wrote, a time you pushed back on a risky change, and how you work with developers who want things fast.
Questions you're likely to get
1.Walk me through what happens when someone types our URL into a browser and hits enter.
Why they ask
It's a quick way to see how many layers you actually understand, from DNS to the load balancer to the app and back.
How to answer
- Start with DNS resolution and caching, then the TCP and TLS handshake
- Explain how the request hits a CDN or load balancer and gets routed to a healthy backend
- Cover the app talking to a database or cache and the response path back
- Point out where you'd look first if any step were slow or failing
2.How do you manage Terraform state on a team, and what goes wrong with it?
Why they ask
State problems cause real outages and ugly weekends. They want to know you've felt that pain.
How to answer
- Remote state in a backend like a cloud storage bucket or Terraform Cloud, with locking turned on
- Split state by environment and by service so one bad apply can't touch everything
- Explain drift, how you detect it with plan, and how you handle an import or a state move
- Mention reviewing plans in pull requests before any apply runs
3.A pod in Kubernetes keeps restarting. How do you figure out why?
Why they ask
This is the most common Kubernetes question in these interviews because it shows whether you've actually operated a cluster.
How to answer
- Check the pod's events and status with kubectl describe, looking for OOMKilled or failed probes
- Read the previous container's logs, not just the current one
- Compare resource requests and limits against real usage in your metrics
- Check recent deploys, config changes and whether a dependency like a database is down
4.Design the infrastructure for a new web service that needs to stay up if one availability zone fails.
Why they ask
It's the core of the design round. They want to see you think about failure first and pick the simple option where it's enough.
How to answer
- Spread app instances across zones behind a load balancer with health checks
- Use a managed database with a standby replica in another zone and tested failover
- Keep secrets in a manager like Vault or the cloud's own service, not in environment files
- Add metrics, logs and alerts that page on user-facing symptoms
- Name the cost trade-off and what you'd skip for a first version
5.Tell me about the worst outage you've been part of.
Why they ask
Everyone in this job has broken something. They want an honest story and proof you learned from it.
How to answer
- Set the scene briefly: what broke and who noticed
- Explain what you did in the first hour and how you communicated with the team and stakeholders
- Name the real root cause, even if it was your change
- Describe the fix that stopped it from happening again, like an alert, a guardrail or a runbook
6.How would you cut our cloud bill without hurting reliability?
Why they ask
Cost is a big part of the job at most companies, and finance will ask the infra team first.
How to answer
- Start by tagging resources so you can see which team or service owns the spend
- Right-size idle or oversized instances using real usage data
- Use reserved capacity or savings plans for steady workloads and spot instances for batch jobs
- Clean up orphaned volumes, old snapshots and forgotten test environments
7.A server can't reach a database on another subnet. Talk me through it.
Why they ask
Networking gaps are the most common weakness in candidates who come from software, so they test it directly.
How to answer
- Confirm DNS resolves the database host to the address you expect
- Test the port with a tool like nc or telnet from the server itself
- Check security groups, network ACLs and route tables in both directions
- Look at the database's own listener settings and allowed client list
8.How do you roll out a risky infrastructure change?
Why they ask
Infra changes have a wide blast radius. They want to hear caution that doesn't turn into paralysis.
How to answer
- Test it in a staging environment that actually matches production
- Roll out gradually, one environment or a small slice of traffic first
- Have a written rollback plan and know how long it takes
- Watch dashboards during the change and tell the affected teams in advance
9.How do you decide what should page someone at night?
Why they ask
Noisy alerting burns out teams. This shows whether you care about your teammates' sleep and your users at the same time.
How to answer
- Page on symptoms users feel, like error rates and latency, not on every high CPU spike
- Send lower-priority alerts to a ticket queue or Slack channel instead
- Every page should have a runbook or a clear first step
- Review alerts after each rotation and delete the ones nobody acted on
10.How do you handle secrets in pipelines and infrastructure code?
Why they ask
Leaked credentials are one of the easiest ways to get breached, and infra engineers hold a lot of keys.
How to answer
- Store them in HashiCorp Vault or the cloud's secrets manager, never in the repo
- Give pipelines short-lived credentials through OIDC or role assumption instead of long-lived keys
- Rotate them and scope each one to the least access it needs
- Scan repos for accidental commits and know what you'd do if one leaked
11.Tell me about a time a developer team wanted something you thought was a bad idea.
Why they ask
You'll say no to people a lot in this job. They want to see you do it without becoming a bottleneck.
How to answer
- Explain what they asked for and why it worried you, with the concrete risk
- Show that you asked what problem they were really trying to solve
- Describe the alternative you offered or the compromise you landed on
- Say how the relationship was afterwards
12.What would you automate first if you joined a team that does everything by hand?
Why they ask
It tests judgment. The right first target is usually the one that's painful and frequent, not the most interesting one.
How to answer
- Spend the first weeks watching where time and mistakes actually go
- Pick something frequent and error-prone, like server builds or user access
- Get it into code and version control so changes are reviewed
- Bring the existing team along instead of replacing their process overnight
Mistakes that sink good candidates
Blaming a past team or a vendor for every outage you mention
Listing tools without describing a single system you ran end to end
Treating on-call as beneath you or asking to be excluded from it
Guessing at a networking answer instead of saying how you'd check
Need more Infrastructure Engineer interviews to prep for?
HeroApply applies to Infrastructure Engineer jobs that match you, every day. 1,449 jobs are open today.