Your goal: design a working retrieval system and threat-model it: who the attacker is, what
they can reach, which controls hold, and what you would do when a test finds a leak.
How this works (about 30 minutes)
- Felix asks which role you are interviewing for (product, engineering or other) and pitches his
follow-ups to it. The question stays the same.
- One design question, taken deep. Ask him about the customer, the documents, the users and the
rules. He hands over what he has, one line at a time. The details are made up.
- Many follow-ups, one curveball, then a trade-off turn from a red-team result: ship, hold or change?
- If you want to sketch, say "can I use the notepad?" and Felix puts it up.
- No hints and no scores during the call. Your report comes after.
What scores well
- A plain pipeline first, with the customer's needs asked before you design
- Access rights applied at retrieval, using the asking person's identity
- A threat model: the attacker, what they control, what the system can reach
- Defenses that sit in code, least privilege, and a place for human approval
- Data care: caches, logs, retention, vendor terms, deletion
- Tests for leaks, honest limits, and a call you can defend under pressure
- Answers that say where they are going and land in a minute or two
Tips
- Take Gen AI Coach · Security first.
- Ask who the users are and how access works today before you draw anything.