Accenture Backend Developer Interview Questions

“I built a proof of concept in two weeks to show a client that their orders could be matched to deliveries without manual work. To move fast, I skipped error handling, used a small sample of data and kept settings in the code. It worked and the client approved the project. I then listed every shortcut and told my lead that the demo was a sketch, not a base. We kept the matching logic, which was proven, and rebuilt the loading, security and error handling properly. I explained this to the client early, so nobody expected production in a month. The result was your result.”

“A requirement said every report must show live data from five systems within a second. Two of those systems updated only at night, so it was impossible as written. I did not send it back with a no. I asked the analyst which decisions the report supported. It turned out only stock levels needed to be fresh, and the rest could be from the previous day. I proposed live stock data and a clear time stamp on the other figures. The analyst took that to the client, who agreed. The result was your result. I learned to ask what a requirement is for.”

“I would design the API around what the app screens need, such as a customer summary, and keep the old systems hidden behind it. For each field I would agree with the client which system owns it, because the same customer often has different addresses in different places. The layer calls the systems in parallel with timeouts. If one is slow, the app still gets the rest, with that section marked unavailable. Data that changes rarely is cached for a short time. Each call carries the customer's identity and is checked, so nobody sees another person's record.”

“I would keep one core that holds the process every country shares, such as creating and approving an invoice. What varies goes behind a clear boundary. Simple differences like tax rates live in configuration with a start and end period. Harder rules get a small module per country that the core calls through one interface. The database stores neutral values, such as times in one zone and amounts with a currency code, and formatting happens when data is shown. Each country has its own test cases written with local experts, and they run before that country's rollout.”

“Proof starts long before the weekend. I would agree the field mapping and the cleaning rules with the client's business owners, because they know which old records are rubbish. Then we rehearse the full load several times with real data and measure how long it takes. After each load, reports compare old and new: record counts per type, totals of balances and spot checks of whole records. Rejected records are listed with reasons. On the real weekend we run the same steps we rehearsed, and business users check a sample and sign off before the new system opens.”

“I would never call the package directly from the code that takes the order. Orders go into a queue, and a sender reads from it at a rate safely below the limit. If the API offers bulk calls, I would group orders to use fewer requests. When the package answers that the limit is reached, the sender waits and tries again with a longer gap each time, and each order carries a unique key so a retry cannot create it twice. I would watch the queue length and the age of the oldest order, and alert when either grows.”

“First I would check the test itself, since an unrealistic test gives a false alarm. If it is fair, I measure where the time goes: usually one query, one lock or one slow outside call. Some fixes are quick, such as an index or moving heavy reports away from the peak hours. I would retest after each change. Within a day or two, I would give my lead and the client a plain picture: what we found, what we fixed, what risk remains and the options. One option must be going live after month-end. Hiding it would be far worse.”

“I like the moment when a new system replaces old ones and a whole business starts working in a new way. Accenture provides consulting and technology services to client companies, and much of that is this kind of large change. It is hard: many teams, a date that matters and data that must arrive complete. I find that pressure focusing. In my last project I worked on loading old records into a new system and found I enjoyed the care it needs. I want to learn how big programmes are planned and how design choices are explained to a client.”