I Built AI Tutors with NotebookLM and Claude to Prepare for Google Interviews

A Claude Cowork routine that scrapes real Google interview questions, builds a NotebookLM study set, and locks the mock interview until you earn it.

Everyone wants to work at Google.
But most people never look at the Google interview questions already sitting in public.
Glassdoor holds 20K+ interview questions for Google.

Glassdoor holds 26,272 Google interview questions. Every one typed by someone who walked out of the room.

Every one of those was posted by a person who walked out of a Google interview and typed what they were asked.

So all you need to do is visit Google careers from here.

Start at Google Careers. Paste the job title you actually want.

Paste your dream job title, for example, AI Engineer, then select the locations near you.

Pick the role and the locations near you. This is the target your team trains against.

Find related interview questions on Glassdoor and start preparing.

So I built an AI team that does exactly that.

Let me show you how it works.

The Claude Cowork Team I Built

Our AI team has three members.

Three steps, three skills, three agents. Nile scouts, Jennifer coaches, Jessy interviews.

• Nile scouts: She goes to Glassdoor through Apify and pulls what candidates reported after their Google interviews. Cleans the pile, labels every question with a topic, and drops the file in my Drive. I read it before anyone builds on it.

• Jennifer coaches: She reads Nile’s file. Glassdoor posts questions and never the answers, so Jennifer researches each topic on the web and loads the material into

  NotebookLM. Then she quizzes me every three days. Free text, graded on a rubric out of 100.

• Jessy interviews: She conducted an interview that lasted 45 minutes, a verdict at the end on Google’s real four-level scale.

Let me show them in action.

How the Interview Prep Team Works

Nile

Nile asks three things before she moves. Token, company, roles.

Nile scrapes Glassdoor and hands you a file.

She asks three things first.

Your Apify token, the company, and the roles.

( Get your apify token here for free; it has a pretty good free quota each month.)

Here is how the file looks, she prepared.

Fifteen questions from twenty records. Read them before anyone builds on them.

Fifteen questions from twenty records. Read them before anyone builds on them.

Jennifer

Glassdoor gives the question and never the answer. Jennifer goes and finds the answer.

Jennifer reads Nile’s file. Glassdoor gives the question and never the answer.

So she researches every topic on the web and loads it into NotebookLM, using NotebookLM CLI.

Next, she sets up a Claude cowork routine, scheduled for every 3 days.

And it’ll finish when you pass 85.

A Cowork routine every three days. It finishes when you pass 85.

Jessy

Locked means locked. Jessy does not take appointments.

Jessy waits until you complete the previous step. Before the interview begins, you need to score at least 85.

Once you hit 85 in three consecutive rounds, Jessy starts the interview.

Forty-five minutes, four timed blocks.

After it, you have 45 minutes.

You can see which part of the interview you’re at.

You always know which block you are in. The verdict lands on Google's four-level scale.

10 Minute Setup

You can set everything up in just 10 minutes. To make it easy, I put all the files into a single folder.

Everything is in one folder. Seven files, three skills.

First, open Claude Code, connect a folder to work in, and paste this.

Set up my interview prep team from this folder.

[Attach to Google Drive Folder]

It holds three Claude skills, Nile, Jennifer and Jessy, split across seven files.

Download every file. Read 0-START-HERE.md first, it says which file goes where.
Create skills/nile, skills/jennifer and skills/jessy/static, then put each file in
its place under the name that file says, dropping the number prefix.

Install what they run on.

  pip install apify-client flask anthropic

Then run Nile.

  python3 skills/nile/scrape_interviews.py --dry-run

She asks for an Apify token, the company I am interviewing at, and which roles. Walk
me through those three answers. Show me the cost, then run her with --limit 20 as a
cheap test before the full scrape.

Show me five of the questions she pulls. If they read like scraped navigation text
rather than real interview questions, stop and tell me.

Then STOP and wait for me to confirm before you build Jennifer.

Here are the files:

What is after the paywall?

The Team Lives in AI Academy

One more thing.

Nile, Jennifer, and Jessy are now agents in the AI Academy, under Agent Teams. Install them from there and skip the folder. The Python files ship inside each agent’s page, and I keep them updated when Glassdoor or Apify changes something.

They sit next to my content team from the One Person Unicorn series. Two teams so far. More are coming.

Agent Teams is an Inner Circle space. Until August 1, annual subscribers get AI Academy access too.