interview-prep
career-ops interview-prep generates a company- and role-specific interview intelligence document, segmented by who is in the room — recruiter, hiring manager, peer-technical, or panel — with likely questions, talking points, and STAR stories drawn from your own profile.
career-ops interview-prep turns a scheduled interview into a working intelligence document. You name a company and role — or hand it a job posting for a role you never formally evaluated — and it researches the hiring process, classifies each round by who runs it (recruiter screen, hiring manager, peer-technical, or panel), and drafts the questions each is likely to ask, each with a draft answer pulled from your own background. Each question is either sourced from public interview reviews found by web search (for example on Glassdoor or Blind) or tagged [inferred from JD]. The result is a saved prep doc, not a chat.
What it does
interview-prep reads the company context (ideally already enriched via deep or oferta), the job description, and your profile, then produces one prep document scoped to the rounds the company's process is reported to include. The doc covers expected question patterns by audience, talking points specific to your background, and pre-written STAR stories you can adapt live. When an evaluation report already exists for the role, that report is authoritative; when you pass a URL for a role you never evaluated, interview-prep fetches the JD itself — structured ATS API first, then Playwright, with a headless fallback — and asks you to paste the posting's text if it is closed.
When to use it
Run interview-prep once a round is on the calendar — it produces a single document with a section per audience the process involves: the recruiter-screen section covers comp-negotiation prep, hiring-manager covers role fit, peer-technical covers architecture and coding, and panel covers leadership signals. career-ops is also told to run it when a role that scored 4.0 or higher moves to Interview in your tracker.
interview-prep is the research step — the intelligence document you study beforehand. It is a different mode from the interactive plan, practice, and debrief interview drills, which rehearse you live and review a round afterwards; they link up (a debrief feeds back into the material interview-prep reads next time). For those drills see the interview modes guide.
What a run looks like
You name the company and role, and the mode returns one interview intel report scoped to the rounds the company's process is reported to include. It researches the process, classifies each round by audience — recruiter screen, hiring manager, peer-technical, or panel — and drafts likely questions, tagging every one with a source citation or an [inferred from JD] label.
/career-ops interview-prep Anthropic "Applied AI Engineer"# Interview Intel: Anthropic — Applied AI Engineer
**Legitimacy:** High Confidence
**Researched:** 2026-05-18
**Sources:** 12 Glassdoor reviews, 4 Blind posts, 3 other
**Audiences covered:** recruiter-screen, hiring-manager, peer-tech
## Process Overview
- Rounds: 4 rounds, ~18 days end-to-end
- Difficulty: 3.5/5 (Glassdoor avg, 12 reviews)
- Known quirks: pair programming, not whiteboard
## Likely Questions — recruiter-screen
- "Why are you looking, and why us?" — 60-90s, anchored to your narrative
- Comp expectation — defer to the band if leverage is thin
- Location / visa / notice period — numbers, not vibesThe report also maps stories from your story-bank.md to each likely question and flags the gaps where you have no story prepared yet.
When to reach for it
Reach for interview-prep once a round is scheduled and you want a working document instead of generic advice. A few things sharpen the output:
- Run
ofertaordeepagainst the company first. The prep reads the evaluation report for archetype, gaps, and matched proof points, so more context yields more specific questions. - Paste the interviewers' names or the schedule when you have them. For a panel, the mode builds a per-interviewer read and drafts one tailored closing question for each slot.
- If you have already named a compensation number for the role, the recruiter-screen section folds it in — career-ops reads your stated figure so the comp-negotiation prep is grounded in what you actually said, not a guess.
- Read the tags: sourced questions cite the review they came from, inferred ones carry
[inferred from JD]. Treat inferred questions as likely, not as what you will be asked.
Related
deep
Generates a deep company research document — funding history, leadership, recent press, product trajectory, comp band signals, red flags. Use it before you apply to a Series B+ company you do not already know well.
Generates a tailored PDF resume against a specific JD using Playwright to render markdown through a clean typographic template. The output is structurally simple — no tables, no columns, no graphics — exactly what ATS parsers handle reliably.