Accenture DevOps Engineer Interview Questions

“At one client, every release waited several days for a manual security review. I asked the security team what they checked. It was a fixed list: scan results, open ports and who had approved the change. I suggested our pipeline produce that evidence in one report for each release. We ran it next to their manual review for a month, and the reports matched their findings. They then agreed that low-risk releases with a clean report could go ahead with a lighter check. High-risk ones kept the full review. After that, your result.”

“I joined a programme three weeks before four development teams were due to start. I spoke with the architects about the technology choices and with the client's IT team about their network and access rules. I could not know everything, so I aimed for a working minimum: a code repository, a pipeline that built and deployed a sample service, and a development and test environment. I wrote a one-page guide for new joiners. On day one the teams could deploy. In the first month I added what they asked for. The outcome was your result.”

“I would separate production from everything else at the account level, and give each team or workload its own space, so a mistake stays small. Identity comes from the client's existing directory, with roles instead of personal keys. Logs from every account go to one protected place. The network connects back to the client's data centre through a shared hub. Rules such as no public storage and approved regions only are enforced by policy. Teams then get templates and a pipeline to create their own resources, so the central team guides without becoming a queue.”

“All environments would be built from the same templates, differing only in size and settings, so a test result means something. For data, real personal details must never reach the lower environments. A masking job runs inside the production boundary and replaces names, addresses and identity numbers with realistic false values. It keeps the same customer linked to the same orders, so the tests still work. For some cases we generate data from scratch. The refresh runs on a schedule, and the client's data owner approves the masking rules. Testers get realistic data and nobody's real details.”

“I would write the plan as a list of steps, each with an owner, an expected duration and a check that proves it worked. Then we rehearse the entire plan in an environment like production, at least twice, and correct the timings with what we learn. The plan has decision points, for example after the data load, with criteria agreed in advance for continuing or turning back. During the weekend one person runs the plan and one channel holds all communication. The old system stays untouched until the client's owner signs off, so going back is always possible.”

“I would first make the bill readable. Without tags that say which team and environment a resource belongs to, nobody can act. Then I sort by size and look at the largest items against their real use. Common findings are test environments running all night, servers far bigger than their load, and storage copies nobody deleted. Those are quick wins. Next come design choices, like data moving between regions. To keep it from returning, each team gets a budget, a monthly view of its own spend and an alert when it jumps. Cost then becomes part of their normal work.”

“I would treat the finding as real until shown otherwise. First I read what the weakness is and check whether our code uses the affected part and whether it can be reached from outside. Then I look for a fixed version of the library. If there is one, we upgrade, run the full test suite and ask for a new scan. If there is none, I propose a mitigation, such as blocking the risky input at the gateway, and I write down the remaining risk. The security team and the client's owner decide. A delay of a day is an acceptable price.”

“I enjoy building the base that lets other teams move fast: a pipeline that works on day one, or an environment anyone can create from a template. Accenture provides consulting and technology services to client companies, and many of them are moving to the cloud or starting large programmes, so that foundation work is needed again and again, each time under different rules. In my last role I set up pipelines for four teams, and seeing their first release go out was the best moment. I want to learn how to design cloud foundations for larger organizations with stricter rules.”