This topic is an interview bank for engineers who use AI tools. Each lesson groups questions by theme and gives what each question tests, a model answer with reasoning, follow-ups and common weak answers.
Lesson | What it covers |
|---|---|
Estimating tokens, context limits, and choosing what stays in context. | |
Worked job cost from example prices, retries, and cost levers. | |
Structured output, examples and reasoning, and debugging a flaky prompt. | |
Checking assistant output and handling secrets and untrusted content. | |
When an agent fits, bounding its loop, and approving side effects. | |
Permissions, human approval, and deciding under uncertainty. |
State example prices as explicit assumptions first. Real prices vary by provider and change, so no memorized figure counts; the formula plus substitution is the answer.