The AI Interview-Prep Coach System
Build an AI coach, mine the real question bank, drill weak spots — then mock with humans
- Difficulty
- Moderate
- Time to result
- ~weeks to results
- Steps
- 5
- Confidence
- 87%
A structured interview-prep system: build a dedicated AI 'coach' project fed with the best frameworks, use an agent to analyze a live question bank so you prioritize what to drill, spin up custom practice apps for weak skills, and demand harsh feedback — while reserving human mocks for the final edge. Use it when preparing for a competitive role interview.
Origin
When Meta reached out, Zevy immediately opened a Claude project as his coach, pulling frameworks from Ben Erez's guest post, and built practice tools — an AI-first reflex he compares to how kids instinctively reach for the phone differently than older generations.
Core principles
- 01Approach every new challenge AI-first
- 02Feed the coach the best available information, not generic prompts
- 03Prioritize practice by what's actually asked, not guesswork
- 04Demand honest feedback; don't let the coach make you feel good
- 05Humans remain irreplaceable for the final, hardest mocks
How to run it
- 1
Build a coach project
Open a dedicated project and feed it the best frameworks and information you resonate with (Zevy used Ben Erez's material); consult it at each phase of prep.
- 2
Mine the real question bank
Use an agent (e.g. Perplexity's Comet browser) to analyze a live, updating bank of real interview questions to learn which are most asked, so you prioritize what to mock.
Pro tip Zevy used Lewis Lin's free always-updating question bank as the source.
- 3
Build custom drill apps
Spin up a small app to drill a specific weak skill — Zevy built a Base44 quiz game for product segmentation and played it on the bus to work.
- 4
Mock with brutal feedback
Tell the coach 'I don't want you to make me feel good — make me as ready as possible,' and have it give pointed feedback; also have it play the perfect candidate so you can learn from an ideal answer.
Pro tip Recording real interviews and having AI critique them closes the feedback loop no one else gives you.
- 5
Do human mocks last
Cold-outreach people on LinkedIn for real human mock interviews — the biggest game changer for a competitive process.
Watch out For elite, highly competitive prep there's no way to get around real human mocks.
In the wild
Zevy built a Claude coach from Ben Erez's frameworks, used Comet to rank the most-asked questions from Lewis Lin's bank, built a Base44 segmentation quiz game, ran AI mocks demanding harsh feedback, and cold-outreached for human mocks.
→ Prepared for and landed the competitive Meta PM role.
Common mistakes
Relying on AI mocks alone
AI mocks take you far, but for a super-competitive process like Meta PM there is no substitute for real human mock interviews.
Is it for you?
Best for
Candidates preparing for a competitive, structured role interview
Not ideal for
Casual or low-stakes interviews where this level of prep is overkill
From the transcript
“I created a project which was my coach uh which I would come and consult what to do at each phase. I would mock interview…”
“you're my coach, and I don't want you to make me feel good. I want you to make me, um, as ready as possible for…”
“the biggest game changer for me was doing uh human mocks. So cold outreaching to people uh on LinkedIn and having them uh do actual…”
From the episode
The non-technical PM’s guide to building with Cursor
Zevi Arnovitz (Meta)