The most stressful part of prep is not knowing what they’ll ask. Skim a few generic interview guides and you get “what’s your biggest weakness?” again — then a pointed, role-specific question on the day catches you cold and all that prep was wasted.
The most accurate question bank is hiding in the job description. What skills the role demands and what experience it values is, nine times out of ten, exactly what gets asked. Hand AI both the JD and your resume and have it predict questions against them — which are high-frequency must-asks, how the interviewer might follow up, and which lines on your resume will get probed — then polish each answer until you can say it out loud. This isn’t luck; it’s putting your prep where it counts.
When to use it
An interview is a day or two out, you’re changing jobs or roles into something unfamiliar, or you’ve got the JD but can’t tell what they’re really testing — that’s when this groundwork pays off.
How to do it
- Open DeepSeek or ChatGPT, paste the full job description, then add your resume or a short background blurb
- Have it predict high-frequency questions from the JD and label what each one is testing
- Pick the most likely few and, question by question, ask for “an approach + a sample answer + possible follow-ups”
- Swap the sample examples for your real experiences, send it back, and ask it to “critique and suggest improvements”
Weak vs strong
The left gives generic questions anyone can google; the right feeds in both the JD and your resume, so the predicted questions track this specific role and this specific you — a totally different hit rate.
Copy-paste prompt
You are a senior interviewer for this role. Here is the job description and my resume:【paste full JD】【paste resume or a short background】. Help me prepare in three steps: 1) against the JD, predict the 10 most likely questions, grouped into “general / role-specific / resume-based,” each labelled with what it tests; 2) take the 3 most critical and give each “an approach + a sample answer + 2 likely follow-ups”; 3) flag which lines on my resume are most likely to be probed so I can prepare.
Worked examples
You get：It surfaces role-specific items like “how do you measure community engagement, which metrics?” and zeroes in on “how will you cover the data-analysis gap?” — exactly the points the JD implies but never spells out.
You get：It teaches the “own the gap + show concrete steps already underway + prove learning ability with a related story” structure, then predicts the follow-up “so what are you learning now, and how far have you got?” — so the second probe won’t catch you out.
Level up
- Tailor to the company: paste the company name, product and recent news, then ask it to “add a few questions tied to this company’s business” — it shows you did your homework
- Self-test in reverse: after predicting, ask it to “play the interviewer and ask me these 3 in a row,” sliding straight into a mock so you actually say your prepared answers aloud
- Global roles: add “this is an English-speaking role, give questions and sample answers in both Chinese and English” to prep content and phrasing together
Common mistakes
- Pasting only the job title, not the full JD: you’ll get generic questions — the JD’s keywords (duties, requirements) are what make predictions accurate
- Memorising the sample answers: interviewers spot a script instantly; always swap in your real experiences and voice — AI only builds the skeleton
- Predicting but never rehearsing: a correct guess isn’t a delivered answer — have it run you through saying the answers aloud to build composure
FAQ
Will AI’s predicted questions actually match the real interview?
Is it safe to paste my resume and the JD into AI?
Pro tip：Predicting is only the first half; rehearsing is what wins — once your answers are ready, ask AI to “play the interviewer and run these questions back to back,” sliding from prediction straight into a mock for double the effect.