Behavioural round: what each level shows, and your own questions

The interviewer compares your stories with a level, as lesson 138 showed for design answers. One true story can be the size of a task. Another can be the size of a system.

A weak candidate tells a small story for a senior role, or makes it sound big. A strong candidate picks the right size and ends with good questions.

Think of carrying water. A child carries a cup, a farmer a bucket, a town a canal. The size of what you carry is the level.

Titles differ between companies. The table is Tradeoff's guide (Senior is SDE-3 in lesson 138). The last column is what Tradeoff's mock asks today.

A scope ladder helps you choose: task, feature, system, several teams. Tell the same story at a bigger step by adding true facts: more people, more time, an unclear goal, or people you influenced.

There are two mistakes. Level-down is a story that is too small for the role. Level-up is making a small story sound big. Follow-ups find it: "Who were the other teams?"

Your questions, and closing

At the end the interviewer asks, "Do you have questions?" Prepare five:

1. What does a good first six months look like?
2. How are technical decisions made here?
3. What is the hardest technical problem the team has now?
4. How does on-call work? (On-call means engineers take turns to answer when the system breaks.)
5. How does the team measure quality?

Avoid salary first, and anything the website says.

To close: thank the interviewer, say one sentence of real interest, and ask about the next step. Tradeoff's mock has a last step for your question.

The double-charge story of lesson 141, at three sizes. All facts are Assumptions (made up). Added facts are in bold.

SDE-1 size. "My lead gave me a task: find why some users were charged twice. I read the logs and found that one payment was sent twice. I wrote a fix and a test, and my lead reviewed them. Duplicate charges fell from 2,300 users to 0. I learned to ask for a review early."

SDE-2 size. "I owned the payment retry feature. I found the cause in the logs. I compared two options: stop retrying, which would fail some payments, or keep retries with a unique number for each payment. I chose the second. I wrote the fix and a test. It adds one database lookup, about 3 ms. Duplicate charges fell from 2,300 users to 0."

Senior size. "Three other services used the same retry code. I wrote a one-page plan, and each team's lead reviewed it. We agreed in 2 weeks, and I built a shared library, code that all four services use. Duplicate charges fell to 0 in all four services. Refund tickets fell from 60 a month to 2. I taught two teammates how to review such changes."

The size grows: a task, a decision between options, then a system and other teams. The predict-box story is SDE-1 size. To make it Senior size you need true facts: other services, agreement, and the measured effect.

• A story at your real level solves a weak signal and survives follow-ups. It costs the chance to look bigger. I would go up one step only if every added fact is true.
• Asking questions solves a weak ending. It costs a minute. I would ask fewer when time is short.

"A bigger story is always better." A story larger than your real work fails at the follow-ups.

"Only senior candidates need a story about people." At every level the interviewer listens to how you work with others. At SDE-1 that is asking for help.

"I have no questions, so I should say nothing." No question can look like no interest.

1. Name the four steps of the scope ladder.
2. Refund tickets fell from 60 a month to 2. What is the fall as a percent?
3. A Senior candidate tells a one-task story. What do you change?