I Found The HARDEST LeetCode Question (79% FAILURE RATE)

Use coupon code "engineered" at nordpass.com/e... to get a free 3-month trial of NordPass Business, no credit card required.
Article: interviewing.i...
🚀 Get promoted in 2025 by taking my FREE 5-Day Promotion Accelerator Challenge: geni.us/9P7CAM
💥 Continue the conversation on my Discord server with like-minded ambitious tech professionals. #accountability is *chef's kiss* and #wins is motivating - / discord
📈Transform your tech career with my free weekly newsletter - alifeengineere...
Top 'Secretly Easy' Hard Problems:
#1million : Reverse Nodes in k-Group - Score: 0.034
#2: Median of Two Sorted Arrays - Score: 0.0298
#3: N-Queens II - Score: 0.0288
#4: N-Queens - Score: 0.0255
#5: Robot Room Cleaner - Score: 0.0192
Top 'Most Efficient for Interview Prep' Hard Problems:
#1: Longest Increasing Path in a Matrix - Score: 1.8094
#2: Trapping Rain Water - Score: 1.6964
#3: The Skyline Problem - Score: 1.5794
#4: Word Break II - Score: 1.5601
#5: Smallest Range Covering Elements from K Lists - Score: 1.1687
Hardest Questions Actually Used in FAANG Interviews:
#1: Partition Array Into Two Arrays to Minimize Sum Difference - Score: 12.3924
#2: Regular Expression Matching - Score: 12.1070
#3: Reverse Pairs - Score: 9.0915
#4: Median of Two Sorted Arrays - Score: 8.2952
#5: Max Points on a Line - Score: 8.0952
01:38 Interviewing.io Article
04:36 Getting the Hard Question Data
06:18 Top 'Secretly Easy' Hard Problems
0852 Video Sponsor
10:10 Most Efficient for Interview Prep' Hard Problems
13:18 Hardest Questions Actually Used in FAANG Interviews
BOOKS THAT CHANGED MY LIFE So Good They Can't Ignore You - geni.us/SoGood This book transformed my life when I read it for the first time nearly 10 years ago. It still guides the way I think about self-development today.
Deep Work - geni.us/ALE-De... The most important currency you have is your undivided attention. This book shows you exactly how to spend this currency in a manner that will lead to head-turning accomplishment.
SYSTEM DESIGN
Designing Data-Intensive Applications geni.us/DataIn... Currently the best reference for contemporary for system design.
System Design Interview (Volume 2) geni.us/System... and geni.us/System... - For interviews, the best references are System Design Interview and System Design Interview Volume 2 by Alex Xu.

[@ALifeEngineered]

Use coupon code "engineered" at nordpass.com/engineered to get a free 3-month trial of NordPass Business, no credit card required. 🚀 Get promoted in 2025 by taking my FREE 5-Day Promotion Accelerator Challenge: geni.us/9P7CAM 💥 Continue the conversation on my Discord server with like-minded ambitious tech professionals. #accountability is **chef's kiss** and #wins is motivating - discord.gg/HFVMbQgRJJ 📈Transform your tech career with my free weekly newsletter - alifeengineered.substack.com/

[@KalyankumarRamaseshan]

Excellent one. It would be cool if you can share the tool you developed to rank the hard problems.

[@randxalthor]

Great breakdown. Getting so much more out of a video than just what's in the title is why I watch. I'll be looking to include the suggested high-value hard problems in my prep.

[@RyanApplegatePhD]

Just read the analysis - one issue is it's not clear if doing more hard questions would actually make you perform better. It's just that people who do hard questions also perform better in interviews, ie there is room for confounders like "ability" or "educational background" which would simultaneously bias people toward doing harder questions rather than easier ones and those things also making you do well in interviews.

[@jarkjek]

Correlation != Causation. Maybe people that solve many hard problems would have been better off solving medium/easy problems. Maybe people that solve hard problems are the type of people that are generally more interested in algorithms, or generally smarter etc and because of that get better jobs and not because harder problems prepared them better. Note: Not saying this is the case. I have as much evidence as you i.e. not sufficient.

[@masteroogway2853]

You're definitely on to something. I would have no difficulties breaking into the industry and whenever i poke around leetcode I don't even bother with easy/medium problems, im not there to improve my syntax recall, i want a challenge to solve

[@phdhgg2404]

it's true. i have 300+ questions but the more questions i do, the less i get from leetcode. it might be cuz i keep choosing medium questions and i dont learn much from it. should try for harder ones.

[@henrysaltandson]

What is a "solving a leetcode problem"? Do you work on each problem until you solve it, never looking at the answer? Do you study a problem for 20 minutes and then look at the solution? If we need to solve 500 problems, what does that mean in terms of time? How much time does an engineer have to anticipate giving themselves before they should be ready to start interviewing?

[@gammalgris2497]

Dont forget problem #1: we build tools for others. Speaking with the user(s) would be good. Organisations make it unnecessarily hard. Leave your ivory tower from time to time. Problem #2: sometimes you have conflicting requirements

[@advithvashist9889]

Thank you! Can always count on you for novel advice :) Don't know why I never considered looking for the most complex questions

[@AnthonyRiceMusic]

Amazing video 🎉 thanks you so much for doing such a good deep dive into what tech companies are asking in interviews and what questions to study to get the most knowledge.

[@toriamfoobar]

Great video thanks, I'd love to see more about mediums cause where I live the faangs are not there, so other companies ask more mediums...

[@mnchester]

Amazing video! Yes, do the same analysis for medium problems. Some medium problems are actually very difficult

[@parikshithegde7477]

Thanks for this superb video steve. Would you be able to share the list of all the problems that you ran the analytics on instead of just 5?

[@okdotpy]

I think there's a balance: Do enough easy/medium questions to cover all the main topics and some variations. For example, start with the most basic binary search problem which is easy. Do a few variations of it (rotated, first/last occurence, upper/lower bound) to understand the variations that exist. Then do this for all the major problem categories. This will expose you to the building blocks most harder problems are based of (I feel like around 1-200 problems should give you decent breadth across all subjects). After that, just focus on hards. If someone is able to pump out 135 hards after only doing 60 easy/mediums, they likely already have a good grasp of the basics. That or they're just looking at the answers and are questionably learning the patterns/material.

[@mjm1986bk]

How do you feel about the idea that it may be very risky pursuing a career at faang? If you get caught up in a layoff you will be rejected by other companies/recruiters for being damaged goods. I am a workaholic so I don't think i'd get booted for performance by I could certainly see getting caught up in bureaucratic bullshit that I have little control over.

[@teccs6353]

It's easier to get out, than to get in. But once you are in, it opens doors. You do not have to stay forever if you do not want to.

[@mjm1986bk]

@@teccs6353 But you could get in and be laid off within the first 6 months. It seems like a gamble. Does it not?

[@ArbiterA1]

When you do scoring for ranking, why do you need to apply constant factors? Applying them doesn’t really change the ranking of the problems

[@ZhiYin]

Hi everyone, I have a question: if I already worked in other companies for 5-10 years, then come to interview at FAANG, will I still be asked these algorithm questions?

[@Zuppified]

Please give us a top 50 list of the easiest hard and most efficient hard 🙏

[@kccccccccccccccc0904]

I’ve solved only 80 and I’m a Staff at FAANG

[@adityantamarapu6239]

Faang ask hard questions because they already have normal people with them. They want to hire those who are atleast better than normal I guess.

[@rhaikh]

The entire premise of these interviews were flawed to begin with - these problems do not relate to the actual job - but now with AI, it seems more irrelevant than ever. They are literally trained on perfect answers and can recall those answers in seconds. It is difficult to detect if a candidate is "cheating" by referencing AI. Indeed referencing an AI is exactly what a prepared and professional engineer would do if faced with a real problem. I guess it's practical today to talk about leetcode, but tomorrow it won't be. It says a lot about this channel and the industry that AI isn't mentioned on this topic.

[@hellowill]

I found leetcode is waste of time. Do something harder like project euler or advent of code. Leetcode is way too streamlined.

[@SBqwerty]

multiplying the acceptance rate by 0.5 doesn't do anything in the formula

[@abudanish196]

Need a video for medium problems for amazon for sde-1

[@petr-heinz]

Nitpick: At kzbin.info/www/bejne/Zom4q6dmeZirg9k - halving a factor in a product doesn't give it smaller weight at all. You're just halving the score. The acceptanceRate shouldn't be multiplied by 0.5

[@dantesbytes]

i want to work for a top tech company

[@PaleBlueDot-c8k]

lovely