Product Engineering interview guide
Every question carries the same frame: the team situation behind it, what it is actually testing, what a strong answer sounds like, and the flags on both sides. The red flags are specific wrong answers a confident engineer really gives — not 'does not know the topic'.
The PM wrote the ticket after a competitor launched one. It has a mock of a heart icon on the product page and a "My wishlist" tab on the account page. Refinement is tomorrow and the team is expected to give a size.
The feature took one engineer about six weeks. The quarterly review slide lists it as "Delivered". Nobody defined success before it started, and the manager now asks each engineer to present one thing they shipped and its impact.
The wholesale customer places large orders by email today and finance keys them in by hand. The requested discount is tiered by quantity and applies only to their account. Checkout already has percentage codes and free-shipping codes.
A behavioural question in a product-engineering interview. The interviewer wants a real example; a hypothetical is allowed if the candidate says so. The follow-ups probe whether the reasoning survives contact with a sceptic.
The team is proposing one-click reorder on the order page. The director has to pick two of five proposals for next quarter. There is no data on repeat purchase frequency by product category in the room, and the question is asked in front of the whole planning group.
Order confirmation URLs look like /orders/104233. Security raised that sequential ids leak order volume to competitors. Separately, the designer wants to test "Pay now" versus "Place order". Both are small tickets on the same board.
Today returns happen by email: support checks the order, emails a label, and finance issues the refund when the parcel arrives. The PM's spec includes a returns portal, label generation via the carrier API, reason codes, exchanges, store credit and automatic refunds.
The store has a flag service that engineers use to merge incomplete work. There are a lot of flags in it, many of them on for everyone for months. The PM has never looked at it.
The team's board moves tickets to "Done" when the PR merges. The PM counts features shipped per sprint. Last month a delivery-date feature was "done" for two weeks before anyone noticed it showed the wrong date for one warehouse.
The designer's goal is "reduce friction". Finance cares about revenue. Support has complained about address errors. The redesign will run behind a flag for a portion of traffic.
The variant lowered the free-shipping threshold. It ran for one week, which included a public holiday. The dashboard shows the variant ahead and marks it "significant". Nobody has looked at margins or returns.
Support forwards a dozen tickets saying the page "spins forever" after paying. The dashboard counts orders created and HTTP 5xx rates. Nothing in either has moved.
Marketing wants more accounts for email campaigns and the PM has agreed to the change. The PM owns the roadmap; you own the checkout code. You have a strong opinion and no data from this store.
A change to the refund job skipped orders paid with a gift card plus a card. The fix is deployed and the backlog processed. The finance lead has to answer questions from the CFO and from customers' banks.
Stock lives in a warehouse system that is queried in batches. The mock is polished and has been shown to the PM. The designer's intent is to reduce the disappointment of items going out of stock at payment.
A config change to the payment timeout caused all card payments to fail. You rolled it back. Support fielded complaints; some customers placed duplicate orders when they retried. It is Monday morning.
Marketing keeps asking for new promotion types. Each one takes weeks and has caused a pricing bug twice this year. The PM has a full roadmap and has previously declined "refactoring sprints".
Reviews shipped years ago. Few customers write them, support spends time moderating spam, and the reviews widget slows the product page. Some old products have many reviews; marketing likes showing star ratings.