Most interview prep online is a mix of leaked questions, hearsay and "Top 75" lists. I wanted the opposite: what each company says about its own interview process, in one place, with practice questions I could trust.
So I read the official hiring pages for Google, Amazon, Microsoft and Meta, and turned them into free guides on Nokku: nokku.payanai.com/interview-prep. No sign-up.
Here's what stood out, then how I built it.
What the companies actually say
Google describes structured interviews: every candidate for a role gets the same questions and the same scoring rubric. Its own guide names role-related knowledge, problem solving and leadership. In 2025 Google also said it's bringing back at least one in-person round, citing AI-assisted cheating in virtual interviews.
Amazon runs a phone screen of 30 to 45 minutes, then a loop of 4 to 6 interviews of 45 to 60 minutes each, including a Bar Raiser: a trained interviewer from outside the hiring team. Every interview checks for its 16 Leadership Principles, and about half of a technical loop is behavioural. Amazon suggests STAR answers of about 2 to 3 minutes.
Microsoft is refreshingly specific: code in your strongest language, no pseudocode, know at least one O(n log n) sort in detail (preferably two), and manage your time in a ~45-minute round. It also says it looks for a growth mindset.
Meta describes a 45-minute tech screen (roughly 5 minutes of intros, 35 of coding, 5 for your questions), then a loop of coding, design and behavioural interviews. The process usually takes about 2 to 3 months. And the big change: Meta's hiring page says technical interviews now include an AI assistant built into the environment.
The common thread across all four: think out loud, get to working code, then improve it. Meta says it plainly: non-optimal but working code beats just an idea.
How I built it
A few decisions mattered more than the code:
- Official sources only. Every guide lists its sources and a "facts checked" date. If something is widely reported but not official (for example Google's "Googleyness" or its hiring committee), the guide says so.
- Original practice questions. No leaked or copied problems. Each one is written in the style of common interview questions, with hints you reveal one at a time and a full solution.
- Every solution is tested. Each one is che version on thousands of random inputs before it ships. Writing the test caught one of my own wrong "expected answers".
- Content as Markdown. Guides are plain `.m rendered with the same components as the site's courses: flashcards, quick checks, step-throughs, and a new "practice question" block.
- Infographics from data. Each guide's "at rs, a timeline, how time splits) is generated from frontmatter. The facts live next to their sources, and the hub builds a side-by-side comparison of all four companies from the same data.
What's inside each guide
- The process as a step-by-step timeline
- What interviewers score you on
- Pattern flashcards (two pointers, sliding winn's Leadership Principles
- Behavioural scenarios with feedback on each choice
- 3 original practice questions with tested sol
- A 4-week prep plan
Try it
👉 nokku.payanai.com/interview-prep
It's free, with no account. I'd love feedback: which company or topic should I add next? And if you've interviewed at one of these recently, does the process match what
Top comments (1)
Keeping the comparison and infographics driven by the same source data is a good choice. For the next topic, I'd add role and region exceptions: which guidance applies to all candidates, versus one technical track or location? A company-wide "facts checked" date can otherwise make a narrow source look universal. Per-fact source links and scope labels would also let you mark one claim stale without making the whole guide look outdated.