Pass your next DevOps interview.
Scenario-based DevOps interview prep. Make the call on a real production incident, defend it to a senior-engineer AI, and get scored against an answer key.
One payment. 12 months of access. No subscription, nothing renews.
What a practice round looks like
You make the call, the senior engineer pushes back, the answer key grades you.
01The decision
3 AM Outage: Roll Back or Fix Forward?
Intermediate - about 12 minutes
It is 3:07 AM. PagerDuty wakes you. Release v412 went out 20 minutes ago, and a teammate in Slack says "I think I can patch it, give me a sec."
Checkout error rate jumped from 0.1% to 9% right after release v412. You can roll back to v411 in about 2 minutes, or push a fix forward with an unknown ETA. What do you do, and why?
You have up to 5 rounds to defend it.
02Defend it
- You
Roll back to v411 now. It is a one-command, 2-minute recovery to a build that ran clean for nine days, and checkout is failing for 9% of users.
- Senior engineer
Your teammate says they can patch it in a sec. Rolling back throws v412 away and means a second deploy later. Why not give them five minutes?
-> Pushes on your weakest point. Never tells you the answer.
- You
Rolling back throws nothing away. The code is still in git, the logs and traces are still there. "Give me a sec" is not an ETA, and every minute is failed checkouts. Restore first, debug v412 after.
- Senior engineer
Fair. You restored service and kept the evidence. That holds up.
03The verdict
- Correctness
- 8/10
- Reasoning
- 8/10
- Handling pushback
- 9/10
- Overall
- 83/100
What you got right
- + Stop the bleeding first
- + Known-good beats unknown ETA
- + Rollback does not block root-cause analysis
To go further next time
- - Check for non-reversible migrations before rollback
Takeaway
During a customer-impacting incident, default to the fastest known-good recovery, which is usually rollback, but verify it is safe (no forward-only migrations) before executing. Restore first, diagnose second. An unknown fix-forward ETA is almost never the right bet while users are failing checkout.
Answer key reviewed by Branko PetricThis one is spoiled now. The other 52 are not.
Read the full scenario ->How is this different from asking ChatGPT?
A general chatbot makes up an answer on the spot. DevOps Ledger grades you against an answer key written before you started.
It cannot give away the answer.
The senior engineer only sees the traps, never the answer key.
It cannot invent praise.
The grader can only credit points that are in the answer key.
It reads for substance.
You get credit for the point you made, not for hitting a keyword.
Which DevOps interview scenarios can I practice?
DevOps interviews are mostly "what would you do" questions. There are 53 here: 16 beginner, 20 intermediate, 17 advanced.
Kubernetes
Terraform
AWS
Security and CI/CD
Interview in two weeks? That is enough time for 20 challenges at 15 minutes each.
Browse all 53 DevOps challenges ->Built by a working DevOps engineer
I'm Branko Petric, a DevOps engineer. I have taken the 2am pages, run the migrations that went sideways, and defended decisions to people more senior than me. What taught me was making a call and getting torn apart on it.
So I built the opposite of a chatbot that agrees with you. I review every scenario and every answer key before it goes live.
How much does DevOps Ledger cost?
$39 once for 12 months of full access. No subscription, nothing renews.
- All 53 challenges, beginner to advanced
- Every new challenge added during your 12 months
- All guides
- The private Discord
Sign in, pay once, start your first challenge. 14-day refund, no questions asked: email branko@devopsledger.com.
DevOps Ledger at a glance
DevOps Ledger is a scenario-based DevOps and SRE interview practice platform: you pick a real production incident, defend your decision in a chat against a senior-engineer AI that pushes back, and get scored against an answer key written before you started.
- Who it is for
- DevOps, SRE and platform engineers preparing for interviews, junior to senior.
- How a session works
- Pick a scenario, take a position, defend it for up to 5 rounds against a senior-engineer AI. Get a verdict (passed, partial or failed) with scores. About 15 minutes.
- Scenarios
- 53 production scenarios (16 beginner, 20 intermediate, 17 advanced) across Kubernetes, Terraform, AWS, Docker, CI/CD, Linux and incident response.
- How it is graded
- Every scenario has an answer key written before your session and reviewed by Branko Petric, a working DevOps engineer. The takeaway you see is copied word for word from that key, never generated during your session. The AI can only credit points in that key.
- Price and access
- $39 once for 12 months of full access. No subscription, nothing renews. 14-day refund, no questions asked: email branko@devopsledger.com.
- Free without paying
- The DevOps guides, the AI skills and the full challenge catalog. Every scenario can be read before you pay.
- Why there is no free trial
- Every round runs a real model twice: once to argue, once to grade. Instead, every scenario is public and a full sample round is on this page.
- Will it help me pass
- It trains making a production call and holding it under pushback. Nothing guarantees an offer, but you walk in having defended real decisions out loud.
- Who built it
- Branko Petric, a working DevOps engineer. He reviews every scenario and answer key before it goes live.
Your next interview will ask you to make a call. Practice making it.
Not ready to pay? Start with the free guides:
- Linux guides
- Docker guides
- Kubernetes guides
- Terraform guides
- AWS guides
- GitHub Actions guides
- All free DevOps guides
Questions? Read the FAQ