We built a crowdsourced interview question database for tech interviews | Hacker News Reader

We built a crowdsourced interview question database for tech interviews

interviewdb.io

85 pointsrascal_rabbit73 comments

Kinda sad that it has come to this.

Maybe I'm naive but I think an interview should be an honest exchange of what you actually know, not what you crammed for before the interview.

In fact, for me it is pretty hard if not impossible to remain genuine to my ability and personal touch to navigate corporate world having to swallow all those keywords and behave properly in front of interviewers. Thinking that hacking/mastering it is often the single way out.

Not sure it should really be "hard if not impossible" to "behave properly."

I took properly to mean conform to a corporate model, rather than something more like 'not childish or rude'

You're probably right :)

It should be. If the interview tests for what you crammed for, it doesn't test for what you'll remember a month onto the job. That's a sub-optimal strategy for the interviewer.

Devil's advocate: there exists a corpus of questions solvable with a small number of techniques after a few weeks of decicated study. When someone has the right brain, these questions are the barrier to a 99th percentile income. What does it say about an interviewee that doesn't bother?

It says they lack the social capital to know what these questions are and how to study them. Is the goal to optimize for people with family and friends already in tech that can pass on the "secret" or to maintain the fig leaf illusion of meritocracy?

That they're unwilling to jump through random arbitrary hoops that are unrelated to their actual work. If a potential employer considers that a negative, what does it say about the employer?

Work is full of arbitrary hoops

I occasionally interview people and that wouldn't really bother me. If you can learn something once you'll be able to re-learn it down the road when needed.

I've found myself in that situation helping my kids with their homework. I might not be able to define L'Hôpital's rule off the top of my head, but it only takes a few minutes of reading to reload it into my brain.

I agree but I companies like Meta/Apple/Amazon hire so much that they probably have full time statisticians just to analyze interview process and map that back to performance. So you have to assume this maps to real world success. Even if you are just testing an employee's preparation, maybe that means something.

It's not probably.

The hiring rubric gets items added/dropped based on what portions are statistically significant when compared to post-hire performance reviews.

There's a lot of research in the public about "Structured Interviews".

When I was an engineer at Amazon, we all just made up our own questions, or passed fun ones around the office. I assume the Bar Raisers probably had an actual rubric, but the rest of us didn't (and we all had interviewing quotas to meet)

How long ago?

I left a while back, but folks still there tell me not much has changed (except they do a lot less hiring these days, so quotas aren't as big a part of the promo package)

The "Big corporation analyzes everything thoroughly and arrives at the most optimal solution" is a ridiculous myth, and I don't think I even need to give examples. Maybe not Apple, but Amazon surely has a lot of embarrassing failures.

>Maybe not Apple, but Amazon surely has a lot of embarrassing failures.

Like pushing a total RTO and not having space or desks for employees then having to delay the RTO.

Interviewing went from something you had to get a job to a completely different skill and even a business.

I struggled early in my career with interviews. Granted this was 2008-2010 era so the job market was not great, especially in a not-quite-urban non-coastal city where I was for a junior developer.

But something did change when I stopped treating interviews as "convince them to like me" / "convince them to hire me" and started treating them as "find out if I want to work (in order) 1) at this company, 2) on this project, 3) with/for this person." I'm sure part of it is just experience, knowing more of the technical stuff, sure, but when you're approaching it honestly as determining whether you give a shit about the person sitting across from you or the glorified Excel spreadsheet you'll be working on it is a lot easier.

It's a big part of why I hate leetcode interviews and consider them a huge red flag especially for non-Tier 1 companies. If your 90 minute interview is 80 minutes of me reversing trees and writing maze solvers then letting me ask 2 questions while you check your email and Slack on your other monitor I am not learning anything from the process and am extremely uninterested in working for you.

What makes those questions ok for Tier-1 companies to ask, in your opinion?

I don't think it's ok, but the compensation and competition are both so high they can get away with it. A senior dev at Google in Manhattan will make more than a general practice physician in the same area. This is for a job that you can do with zero credentialing, no degree required, for half the hours and none of the liability. The competition is insane.

Me being on my high horse is not going to change Facebook's hiring practices. But if I get asked to interview somewhere local where they may only be interviewing a half dozen people for a role over a few months, and I stop it early because they start throwing out LeetCode hards, that could be a non-trivial signal for them.

Salary

At a tier 1 tech company the work will be ads/tracking on a global scale. You know that before you enter the door, and if you didn’t like it you wouldn’t be interviewing there.

So much less need for time to decide if you’re interested in working there.

This is a great point. Nobody interviewing at Facebook is unsure if they want to work at Facebook.

The problems FAANG companies and an exceptional group of frontier AI companies are solving are leagues different to the problems regular startups think they are solving.

So it only makes sense for those Tier-1 companies to be asking questions that are related to the a vast amount of problems they face on a regular basis.

Your startup is not Google.

Tier 1 is default alive and thriving. They don't need the smartest people. They need people who do not screw up. Someone who's willing to spend 3-6 months cramming interview questions is also someone who is unlikely to drop the ball.

Startups are the opposite. They are default dead. If they hire good enough people, they die. They need exceptional, hungry people. The kind of people who abhor busywork. Startups will pull much lower quality on average and paying higher rates drastically increases the odds of death. So their hope is to get lucky on an exceptional junior. They're also more willing to fire fast instead of doing some odd layoff down the line.

Basically Tier 1 will bias towards minimal false positives, and startups will bias towards minimal false negatives. Most companies are somewhere in the middle, default dead in 30 years or so.

Tier 1 also has a huge funnel and they hire lots of people so they need it. A lot of junk gets into the funnel because of the sheer size of it. They need that 1 out of 300, so they just layer enough filters to hit that rate.

Tier 2 companies who need to filter 1 out of 50 should not be adopting Tier 1 practices. Startups probably have like 10 applicants and 6 of them can't do FizzBuzz. But they still need to filter, even if sometimes they end up with 0.

> Someone who's willing to spend 3-6 months cramming interview questions is also someone who is unlikely to drop the ball.

I don't believe this at all, they're unlikely to not comply and execute arbitrary commands. Which at least in my experience is orthogonal to whether they're able to actually problem solve or build good things.

This sounds good but I don't think it lines up with reality at parts.

I don't think there's any correlation between "willing to spend 3 months cramming LC Hards" and "doesn't screw up on the job." The speed at which you can figure out time complexity in your head, or how clean your maze solver is on the first try has nothing to do with your skill or ability in managing services in AWS or implementing a Figma design or making sure you're utilizing cache correctly.

> Most companies are somewhere in the middle, default dead in 30 years or so.

Uh, no? What's the evidence supporting this? I haven't seen any organizational literature or research suggesting you have "Tier 1 tech" on one extreme, "Tech startups" on the other extreme, and "every other company" in the middle.

I think the general thrust of your point with regard specifically to Tier 1 tech is spot on - keep on layering more and more filters until you go from 500 applicants to a handful, then make offers to those people. My sibling comment from yesterday says basically the same thing, that it's not "ok," but TC and competition are so high they can get away with it. Home Depot or Kaiser Permanente or JP Morgan can't, despite having their own set of unique and pretty interesting technical problems.

The startup stuff doesn't line up with my experience though.

I've had two "Online Assessments," as they're called, in the past couple of weeks and they're even weirder than that now. The request access to my desktop, webcam, microphone, akin to a proctored exam. They don't let you use your own editor, or even leave the window. There is no one there to interact with, ask clarifying questions, etc. It then plays everything back with keystrokes sync'd to video/audio for the reviewer.

If I have to go through that bullshit, I would at least like the opportunity to use it to show how I think, ask clarifying questions even if it is just to show understanding of edge cases, create some small talk, etc.

I think that would be a hard pass for me.

Why would I allow some rando company’s corporate spyware onto my personal computer?

If you’re that worried about me using AI, maybe the interview should be on-site. And maybe it should be relevant to the work I’d be doing, not solving a bunch of leetcode questions.

Or maybe the dystopian future of tech interviews just involves a burner laptop tethered to a burner cell plan.

The future is absolutely flipping fantastic.

If I had that leverage, I certainly would hard-pass, but the market just isn't there right now. I have hands-on dev experience (and the education) in different roles, but none of those roles are solely as a "software developer," so just getting the opportunity to do these goofy things is almost necessary just to get a seat at the table so to speak to begin talking about my background and KSAs. One thing to note is it is all perms given through the browser.

The most recent example for me was even worse- the position was for a web dev asking for experience in C#/.NET. The integrated IDE had all of the usual spyware did not actually compile anything or run any example checks. Then it proceeded, without any brief at all, to ask 7 more 'optional' gotcha-like questions on C pointers and nuances.

It's a terrible time to want to pivot into dev full-time, from the lense of being in the job market anyways.

> Maybe I'm naive but I think an interview should be an honest exchange of what you actually know, not what you crammed for before the interview.

All the way back in ~2012, Amazon's own recruiters were recommending candidates cram "Cracking the Code Interview" (to the surprise of the actual interviewers, who didn't want you doing that). This is the way things have been for a long time now.

Maybe it's time to question it. I don't think there's anything wrong with the now traditional whiteboard interview as such, but the notion that the same questions are all just as relevant to disparate teams is highly questionable. Technical questions should be based on the actual codebases or systems, IMO. I would also question the Hogwarts style approach of interviewing the candidate as a generic dev and then sorting them to the right team.

It hasn't been an honest exchange of what you actually know in over 20 years.

When you're interviewing junior developers, you have no standard measure, they don't have enough actual experience to speak to. So before a leetcode and CTCI, you'd either have junior developers who went to a good university and were able to get an internship via that network, or they made some great open source contribution or project on the side as a student

Indeed. This has turned into a show about nothing. Recreational and competitive programming sites incentivize candidates for speed and one-upping. With such a group in the candidate set, companies rose up to the challenge of defeating them at their game by picking even harder questions. With the rise of AI, candidates don't even have to know or read the question.

And eye for eye leaves the whole world blind. I don't think there are any winners in this war.

As a hiring manager or company owner (I wore both hats) I developed a routine during the interview that basically leads to a breaking point.

I want to either drill down your bullet points to find out if you're actually an expert in XYZ as the resume claims, or work with you on a coding problem that's simple to understand yet more complex than leetcode challenges.

I don't expect you to ace, but I want to check if you're honest and willing to admit your limitations - my philosophy is that I hire humans, not compilers, no one knows how everything, and I'd rather have someone who's fast finding correct answers on the internet than someone who had memorized a book.

That's how I like to do it. Its been so long that I don't even remember when I myself was treated this way during an interview.

You're not filtering on honesty, but desperation. The desperate will lie because of the higher stakes for themselves.

I dunno. If your immediate response is to lie rather than admit that you don't know something, it increases the probability that this is the person's default response when in such a situation. You don't want that person working for you if you depend on them for critical stuff, like many development or similar positions.

If you've ever worked with someone like this, you'd know that it's frustrating and unpleasant mid-to-long-term. Someone breaks prod and refuses to admit it unless given a commit as evidence and a confrontation rather than owning up and fixing/helping fix it? Pretends to know everything then produces terrible work that sucks more hours out of other devs? Nah.

Admitting fault and lack of knowledge is important, and it's different to ignorance.

If you're desparate, actually know your stuff, and still lie, that's on you IMO. If you don't know your stuff and you lie, you probably wouldn't have made it anyway.

That presumes I can't catch people lying to me?

When prepping for tech interviews, my go-to strategy has always been to scout online for any interview questions and insights people have shared. However, this process is incredibly time-consuming. Information is scattered across multiple platforms: Leetcode Discuss, Glassdoor, 1point3acres (IYKYK) and often buried under low-quality or vague posts. Many threads are just people asking for questions rather than sharing them (Leetcode Discuss I'm looking at you), and even when questions are shared, they often lack enough detail to be truly useful. Typically you have to go through a lot of posts before landing on a good one. To solve this, we built interviewdb.io: a crowdsourced database where people can share real interview questions they've encountered. To incentivize contributions, we’ve implemented a points-based system: you earn points by submitting questions and use them to unlock others.

You might ask: Why Not Just Use Leetcode Tagged? We see InterviewDB as a complement to Leetcode Tagged, not a replacement. Here’s why: 1. Many companies are poorly covered on Leetcode Tagged, with little to no data. 2. No context on whether a question appeared in an OA, phone screen, or onsite. 3. Companies often tweak Leetcode questions, making them harder to recognize under pressure. Seeing real-world phrasing helps. 4. Leetcode doesn’t cover system design or company-specific interview formats/processes.

We launched recently and already have ~5,000 users contributing and benefiting. The more people join, the more valuable this will be for everyone. Would love to hear your thoughts and feedback!

For those who work in the dark matter of the industry, there are no real resources because most places don't even have "question banks". When an interview request pops up on my calendar, I just go out and find a random question - it's totally disposable and unlikely to be used again, and we don't interview enough for someone to compile enough information on us.

None of the places I've interviewed at in the last 5 years are on this list. One of them has a single question reported on Glassdoor and wasn't similar to what I was asked.

Bigger companies do have question banks. Finding a random question at interview time makes it (1) difficult to compare across candidates; (2) makes the interviewer work too hard since the interviewer hasn't first thought about the question and makes the hints given to candidates less useful.

And Google, being the maker of a search engine, was good at detecting leaks of their questions.

I work for a large company that hires a lot of developers. If you're in the US you've heard of us and there's maybe a one-in-three chance you're our customer. We are not "sexy" though. I have seen question banks circulate occasionally but when it's time to actually interview nobody can ever remember where the question bank is. Usually when I've been on a team that's actively hiring we make up a bank of questions for that particular round of hiring, which helps us in comparing those candidates that are directly competing against each other. But nothing at scale.

It's definitely reused again.

And there are definitely question banks. I saw my team go through (the company recommended lists) in a meeting and we selected some questions

Yes, question banks exist. I'm talking about the company I work for and in fact all the companies I have ever worked for. We did not use question banks.

Your claim was most companies though but to be fair, you are more accurate. Most of The big tech companies that are viewed heavily on levels.fyi definitely has question banks.

> For those who work in the dark matter of the industry

Can you be more specific about what this means?

Most software engineers work at companies with names the average person wouldn't recognize. Like, some random B2B company in Ohio that manages corporate gift cards, with 100 total employees. Multiply that by a million. These companies may come up on Glassdoor, but there won't be enough posts to get any kind of consistent pattern on how they run their interviews.

Every day I thank god I finished my career in software before Leetcode-driven interviewing became so popular.

Curious, did you retire or move to a different field? If the latter, what are you doing now?

Yet, these tech interview tests and questions have already been gamed to death by very clever folks like the creator of InterviewCoder: [0] who is now making more than $100K a month out of it.

But do you know what? Good. It is for the best.

This also signals the definitive end of Leetcode, Hackerrank and all the other useless online assessment tools. It is now a completely ineffective tool thanks to AI.

It also is an example that fits the Y Combinator definition to find those who are "hacking a system to your advantage" instead of a "computer system".

Whoever built this destroyed the online assessment system with all its participants in it.

If you're still using online assessment tools like Leetcode, it's now time to find a much more rigorous method of interviewing candidates and also bring on-site interviews back.

[0] https://www.interviewcoder.co/

Do you really need this?

It's usually enough to think as HR/recruiter. Aka open favorite LLM chatbot and let it generate these questions. Because that's what they will do.

Maybe after this “takes off”, they can explore some other domains.

“We built a crowdsourced examiner question database for pilot exams.”

“We built a crowdsourced set of best responses for marginalized peoples who come in contact with hostile police officers.”

So on, so forth.

Then they can franchise it into a set of “exchanges”.

They do actually have crowdsourced databases for pilot exams now that the question bank is not longer published by the FAA

Might want to include a DMCA takedown address, so that you don't get them sent to your upstream provider. I regularly submit takedown requests to other sites when I see my interview questions posted there. I'm sure others do as well.

Is an interview question copyrightable?

I am sure that if someone is uploading the specific wording of an interview question, then that would potentially be a copyright violation. But if they're just describing the problem in their own words (for instance "I was asked to design a system for selling movie tickets") then there shouldn't be a copyright issue.

Not sure what the legal technicalities are, but one of the questions I saw on the site was just a picture of a the user's screen with the question on it, so it's definitely more the former than the latter.

Yes, the bar for copyright protection on written text is very low. And I've never not seen them just plainly copy/pasted.

If someone rephrased them I really wouldn't have an issue with that, because the phrasing is fairly important to the particular questions I'm asking. (and the doofuses using these sites probably would miss any nuance when rephrasing them anyway)

Blind, Leetcode and Chinese sites like 1point3acres have a ton of interview questions.

“Technical review by experienced developers Our team of engineers with experience at top tech companies reviews each question for technical accuracy, relevance, and clarity.”

Which engineers which top companies? It’s all very ambitious.

I think hiring process is broken, many capable candidates are eliminated until hiring someone. Time and money is lost, hearts are broken :)

Maybe there is a big opportunity for a startup, that will revolutionize hiring.

I'm on mobile so maybe the desktop experience is better, but it would be nice to search by job title rather than company.

So safe to say these questions wont be showing up on any upcoming interviews.

Just curious - where is that hero image from? It feels familiar

How about HackerRank assesments?

Hot take - these types of interviews are caused by "elite overproduction" - there are too many very qualified and capable programmers, so the end result is a process that is selecting for the most agreeable and obedient personalities who agree to put up with these gauntlets.

In the dotcom boom, my impression was that (pre-existing Microsoft interviews aside) current techbro interview baggery was caused by a bunch of affluent people rushing into the field, and (for various reasons) seeking to erect barriers to entry for people who are not them.

But now most people working implicitly assume that the interviews are supposed to be some BS test rituals, rather than a collegial dialogue. So now we're just beating anyone who tries to climb the ladder to get the bananas.

I think you're right that there's also something to the obedient corporate drone theory, for some very large and stodgy companies, in the exact flavor of interview nonsense that they do. But at the bulk of tech companies, I think a lot of hiring managers just want people who will be productive and professional, and these interview rituals don't seem to be good indicators of that.

It may also be that "elite overproduction" is also going on, depending on how you define the terms. Or just oversupply of people who meet the (as XKCD put it) "our entire field is bad at what we do" bar for competence. But in any case, I don't think that overproduction is the cause of the nonsense interviews, but maybe overproduction permits the process to be ridiculous.

If by “elite” you call everyone who can write some js then sure.

If you mean developers that know at least something about software- then this is definitely not the case.

I mean, if theres people that need this i won't judge - but im in tech for ~20 years now and its really not hard to formulate interview questions if you know what the target position is and also have a little of social skills.

This gamified payment model is not my cup of tea.

The whole vibe of this site seems like it’s more interested in growing user count and making money rather than helping people.

I also feel like “crowdsourced” is a euphemism for “stolen content.”

Sure, it’s a capitalist world and I’m not against starting a business, but this feels like something supremely lazy, like the author is hoping to make money without any of the effort of starting a real business. Just be a middleman to other people’s work.

Idk, good luck competing with AI chatbots.

Maybe the real business model is to sell the data to the LLM trainers.

Unless said LLM trainers figure out some way of downloading all the data without paying...

Reply on news.ycombinator.com