Elman Huseynov
Tech Interviews

The Complete Guide to FAANG Technical Interviews

7 min read

Episode 10. The stage that decides is the one you are not in

The FAANG interview process has 5 stages.

A recruiter screen, a technical phone screen, an onsite loop of 4 or 5 rounds. Then a debrief, and team match.

You are only judged in the room for 2 of them.

The decision happens in the debrief, from written notes, by people who never met you.

Nobody rejects you. A committee does, from what the interviewer wrote down.

Most candidates optimise for the hour instead.

I have sat on both sides of this. More than 100 interviews run, over 10 engineers hired.

Hiring processes defined across 27 markets at Zalando. Plus my own loops at Google, Amazon, Meta and Shopify.

Here is the whole machine.

Who actually decides whether you get the offer?

A committee decides, and in most cases it never met you. Your interviewers do not vote.

They write up what happened and present it, and the group reads those notes.

So the onsite is the stage everyone prepares for. The debrief is the stage that decides.

Stage Length Who decides
Recruiter screen 20-30 min Recruiter
Technical phone screen 45-60 min 1 engineer
Onsite loop 4-5 hours Nobody yet
Debrief or hiring committee 30-60 min A group, from written notes
Team match and offer days to weeks Hiring manager and recruiter

The strength of that written evidence depends on how easy you made their job.

Stage 1: what is the recruiter screen for?

It is a filter, and it takes 20 to 30 minutes.

The recruiter checks 3 things. That you are real. That you are available.

And that you are roughly at the right level.

No technical judgement happens here. Two mistakes cost people the stage, and both of them are about numbers.

Giving a salary number first. You do not have to. “I would rather understand the level and scope first, then talk numbers.”

That is a normal answer, and recruiters hear it daily. If they push, ask for their band.

Being vague about level. Say “senior or staff, I am flexible” and you get the lower one. Pick the level you want and say it.

Ask the recruiter what the loop looks like. They will tell you. This is free information that most candidates never request.

Stage 2: what does the phone screen test?

One engineer, one or 2 problems, 45 to 60 minutes in a shared editor. The bar is lower than the onsite and the format is tighter.

You get through it by making your reasoning visible.

Restate the problem. Ask about input size. Cost the brute force.

Then write the real solution.

You have 45 minutes and roughly 35 of them are yours. Use them in this order, out loud:

  1. Restate the problem in your own words.
  2. Ask about input size, edge cases and what “valid” means.
  3. Say the brute-force approach and its cost.
  4. Say why you are not doing that.
  5. Write the real solution.
  6. Walk 1 example through it by hand.

Step 6 catches more bugs than staring at the code. Interviewers notice when you find your own bug.

Stage 3: the onsite

4 to 5 rounds in a day, or split across 2 days. Typically 2 coding, 1 system design, 1 behavioural, sometimes 1 domain deep-dive.

The coding rounds

You are tested on what you do when the problem is new to you. Learn about 15 patterns rather than 300 problems.

2 pointers, sliding window, binary search on the answer. BFS and DFS.

Topological sort, union find, heaps, intervals, prefix sums, backtracking.

Dynamic programming on 1 and 2 dimensions. Tries, monotonic stack, greedy.

Almost every question is one of those wearing a costume.

Naming the pattern is separate from writing the code. It collapses first under pressure.

I built AlgoPath to train it. Each structure is taught through the system it runs in.

The problems solve in 6 languages. The pattern trainer hands you a problem with no heading.

The 8-week version is in how to prepare for a FAANG interview in 8 weeks.

The system design round

This is where senior candidates lose offers most often. The reason is nearly always the same.

They start designing before they know what they are designing.

Spend the first 8 minutes on requirements.

Who uses this, how many, what must never break. What may be stale, what is out of scope.

Write the numbers on the board.

Then design to those numbers.

A design right for 1,000 users and wrong for 100 million is a bad answer.

So is the reverse, and the reverse is far more common. People rehearse the big one.

The behavioural round

Companies weight this more heavily than candidates expect. At the staff level it can decide the whole loop.

Write 8 stories before the interview.

A conflict. A failure. A time you changed your mind.

A time you shipped something hard. A time you disagreed, then committed.

A time you mentored someone. A time you cut scope.

And a time you were wrong in public.

Each story needs the situation in 2 sentences, and what you specifically did.

Then the outcome with a number, and what you would do differently.

That last part is the one people skip and the one that separates answers.

Stage 4: what happens in the debrief?

Your interviewers write up their notes and meet. At some companies a committee that never met you reads only the written record.

So the interview is a writing exercise as much as a thinking one.

If the interviewer cannot rebuild your reasoning from their notes, it does not exist.

Things that make you easy to write up:

  • You said your assumptions out loud, so they got recorded.
  • You named the tradeoff you were making, not just the choice.
  • You gave a number: latency, cost, complexity, team size, timeline.
  • You finished a thought before starting the next one.

Things that make you hard to write up. Long silences. Jumping between ideas.

Answering a different question than the one asked.

And “it depends”, without saying what it depends on.

Stage 5: team match and offer

At some companies you are hired by the company and then matched to a team. At others the team hires you directly.

Ask which it is. It changes your leverage.

Hired by the company? You can turn down a team without losing the offer.

If the team hires you, the offer and the team are the same decision.

On the number itself: the first offer is rarely the last.

What is negotiable goes well beyond base salary. The sign-on bonus is usually the most flexible line.

What actually gets people rejected?

From inside the debriefs: rejections almost never say the candidate did not know the algorithm.

They say the candidate never clarified requirements, or jumped to code.

Or could not explain the choice. Or went quiet for 4 minutes.

Every one is fixable in a week. None of them are on LeetCode.

Each of those failures is a writing failure before it is a thinking failure. The interviewer had nothing to put in the notes.

Does a referral improve your odds?

A referral gets your CV read and that is all it does. It does not lower the bar and it does not survive a weak loop.

Ask people who have seen your work.

A stranger who refers you is guessing. Hiring managers can tell the 2 apart.

More on this in the hidden job market is mostly just referrals.

Common questions

How long does the whole FAANG interview process take?

Six to 10 weeks is typical, from recruiter screen to signed offer. The gaps between stages are longer than the stages. Amazon and Meta move fastest. Google is slowest, because of the committee step. Running several loops? Ask each recruiter to align the timing early.

Can you fail one round and still get the offer?

Yes, and it happens often. One weak round out of 5 is survivable when the others are strong. And when the interviewer wrote up what you did well. 2 weak rounds usually ends it. At staff level a weak system design round ends it alone. That round carries the most weight.

How long should you wait before reapplying after a rejection?

Most of these companies set a 6 or 12 month cooldown. The recruiter will tell you which applies. The clock starts at the rejection, not the application. Use the time on the thing that failed. Reapplying with the same preparation produces the same debrief notes.

Does the level you interview for change the questions?

The coding questions barely change. Everything else does. At staff the system design round gets more ambiguous. The behavioural round asks about influence without authority. The bar for the debrief write-up is higher. Ask the recruiter which level you are targeting first.

The short version

The loop has 5 stages. The one that decides is the one you are not in.

Ask the recruiter what the format is. Learn 15 patterns, not 300 problems.

Spend the first 8 minutes of system design on requirements. Write 8 behavioural stories with numbers in them.

Say your reasoning out loud so it survives into the notes.

If you want a scored run before the real one, that is what I do.

A mock interview is 60 minutes, 100 euro.

You get the same scorecard a real panel would fill in.