roadmap
- 01
- 02
- 03
- 04
- 05
- 06
- 07
- 08
Prepare for threat modeling interview questions on what can go wrong in a design before release. Covers data-flow diagrams, trust boundaries, STRIDE, attack trees, risk rating and SDLC practice.
232 of 232 questions in your plan: 0 known, 0 in progress, 0 due, 0 skipped, 232 new. 0 cut, outside your plan.
Application and product security engineers usually meet threat modeling as a round of its own. The interviewer names an ordinary system, such as a password reset flow, a file upload service or a mobile client behind an API, and asks what could go wrong with it. The grade goes to the method more than to the threats found: a candidate who first asks what the system protects and from whom, sketches its data flows and marks where trust changes before naming any attack does better than one who recites a vulnerability list at an empty whiteboard. The follow-ups test judgment: why a boundary belongs between the API and the queue it writes to, which threats ship as accepted risk, and who signs that acceptance. DevSecOps and platform interviews come at the subject from the delivery side. They ask less about one model and more about the practice around it: which changes should trigger a review, how a model stays true once the architecture drifts away from it, and what a tool can generate from code or a diagram that still needs a person in the room. Security architects and senior engineers who run design reviews get a mix of both, often with a comparison of methods, such as when a privacy method or an attacker-centric one fits a system better than STRIDE. For the loop around this subject, the [Cyber Security Expert roadmap](/roadmaps/cyber-security) sets it beside the network and host questions a security role is also asked, and the [DevSecOps Engineer roadmap](/roadmaps/devsecops) beside pipelines and cloud controls.
The plan above is for answering; the reading lives in the [Threat Modeling hub's primer](/topics/sec-threat-modeling), which shows how the parts feed one another, from the diagram to the threat list to the risk decision, a chain the questions take for granted. Reopen it when your answers start naming threats without saying where in the system they sit. At junior level one skill carries most rounds: drawing a data-flow diagram and [placing its trust boundaries](/topics/sec-threat-modeling-dfd-trust-boundaries-boundary-placement) where privilege or ownership actually changes, not around every box. With that in hand, [STRIDE](/topics/sec-threat-modeling-stride) becomes a check you apply to each element of the diagram rather than six words to memorize. Past junior level, the early hours shift to [decision and ownership](/topics/sec-threat-modeling-risk-rating-decisions), because the follow-ups rarely stop at the threat list: they ask which threat gets a fix, which is accepted, who owns that call and when it is reopened. If the interview is days away, let the filters do the triage: importance raised to must know, level narrowed to the seniority the job posting names. The time saved belongs at a whiteboard, sketching a system you have worked on and running each question against the sketch, since the exercise in the room starts from a diagram too.
Application and product security engineers usually meet threat modeling as a round of its own. The interviewer names an ordinary system, such as a password reset flow, a file upload service or a mobile client behind an API, and asks what could go wrong with it. The grade goes to the method more than to the threats found: a candidate who first asks what the system protects and from whom, sketches its data flows and marks where trust changes before naming any attack does better than one who recites a vulnerability list at an empty whiteboard. The follow-ups test judgment: why a boundary belongs between the API and the queue it writes to, which threats ship as accepted risk, and who signs that acceptance.
DevSecOps and platform interviews come at the subject from the delivery side. They ask less about one model and more about the practice around it: which changes should trigger a review, how a model stays true once the architecture drifts away from it, and what a tool can generate from code or a diagram that still needs a person in the room. Security architects and senior engineers who run design reviews get a mix of both, often with a comparison of methods, such as when a privacy method or an attacker-centric one fits a system better than STRIDE.
For the loop around this subject, the Cyber Security Expert roadmap sets it beside the network and host questions a security role is also asked, and the DevSecOps Engineer roadmap beside pipelines and cloud controls.
The plan above is for answering; the reading lives in the Threat Modeling hub's primer, which shows how the parts feed one another, from the diagram to the threat list to the risk decision, a chain the questions take for granted. Reopen it when your answers start naming threats without saying where in the system they sit.
At junior level one skill carries most rounds: drawing a data-flow diagram and placing its trust boundaries where privilege or ownership actually changes, not around every box. With that in hand, STRIDE becomes a check you apply to each element of the diagram rather than six words to memorize. Past junior level, the early hours shift to decision and ownership, because the follow-ups rarely stop at the threat list: they ask which threat gets a fix, which is accepted, who owns that call and when it is reopened.
If the interview is days away, let the filters do the triage: importance raised to must know, level narrowed to the seniority the job posting names. The time saved belongs at a whiteboard, sketching a system you have worked on and running each question against the sketch, since the exercise in the room starts from a diagram too.