Coding Interviews in 2026

It's 2026. What is going on with coding interviews and maybe coding in general? Well, the bad news is that honestly I have no idea. But the good news is that nobody really has any idea what's going on anymore. And that's half the fun. So, let's talk about it. Are people still doing DSA coding interviews? Should you still study DSA? Should you still learn how to code? Or should you just learn how to prompt? Maybe learn all these different concepts related to that. Agents, vibe coding, AGI is just around the corner. And you know, if you go on Twitter, everybody's just writing all of their code with AI tools nowadays. You get the sense that there's no value in technical skills anymore. or at least that's the perception but when you do look a little bit more closely so for example to use a concrete example anthropic is really big like their CEO is saying well he said last year coding 90% of code is going to be written by AI tools and quite honestly I thought that was full of crap but practically we are at that point so why is anthropic still hiring engineers developers And not just developers, but they're actually hiring experts. They hire, they aqua hired Jared from Bun, the very competent developer, or you could say he is now the adult in the room working on Claude Code, making uh maybe cleaning up a bit of the mess that was made when you go fast and break things or nowadays when you vibe code, which you know, maybe the term I would use is go fast, eat ass because, okay, anyway. ways. But so there's a bit of a contradiction right now going back to interviews, Anthropic is actually asking DSA questions. What? You wouldn't believe that, but they actually are. And I had one of the developers of CL one of the people working on cloud code reply to a tweet of mine saying they used my website to prepare for the coding interview at Anthropic. Now, how does that make any sense? It's like everywhere you look, there's contradictions. And actually, Open AI is also asking DSA questions. Can you believe that? So, it seems like some things aren't uh changing, but you look in other corners, even in big tech, you see meta. You see they're exploring something new. They're exploring their AI assisted coding interview. And I'm actually going to be adding some resources explaining these AI assisted coding interviews cuz I think there's not really a lot of resources that kind of aggregate all the information in one place. So expect a link below. As soon as I have that, I'll probably link it below this video. And these types of interviews, these give the impression that DSA is dead. No more DSA. No more dynamic programming or crazy graph traversal. So, you hate DSA. You're probably happy, right? Well, I really don't want to break the news to you. Um, but it turns out, well, let's use Meta as the example. It turns out that Meta's assist AI assisted coding interview, first of all, it's not easy. It's actually harder than the typical DSA nick code style whatever question that you would get. You still need knowledge of algorithms, but now you also get the chance to use an AI tool. You typically like some even some of the state-of-the-art models, but now that you have the AI, the bar is higher. So if you were hoping that once leak code dies that well it's going to be easier now you can finally get your foot in the door. Unfortunately that's not the case and I think it was kind of obvious that was going to happen because the reason interviews are difficult is there's such a wide distribution of people and a company is typically going to go for like the best people or at the very least eliminate the false positives. Typically, it's about eliminating those. And so you know in those interviews things that come up is the concept of unit tests. Hey, same thing in DSA interviews. Okay. Well, another thing that comes up is the performance of your code. Same thing in DSA interviews. You need to know certain textbook algorithms like maybe graph algorithms or you know array traversals, things like that, hashmaps. Okay, sounds

like a DSA interview. You need to be able to communicate. explain why you're using a piece of code. You can't just take the AI generated code and copy and paste it in there. And uh I think that's really important because I'm going to step back for a second and talk about something that's not entirely related to coding interviews. I'm actually going to be talking about well what happens after the interview because I've uh you know working on neat code. I've been hiring people. I've been, you know, ever since the Opus 4. 5 model came out back in November of last year, there has been an inflection point. But in the past, people might have gotten the perception I'm an AI hater. Well, I was only a hater because I felt that the capabilities were being overstated. But right now, actually, I think I mean, the capabilities sometimes are still overstated, but the capabilities are way better than I would have anticipated. But the result is not quite what you would have expected because first of all as I'm working on neat code I'm actually able to move so much more quickly than I could in the past. But at the same time when I hire somebody they're also able to move quickly because now they don't need me to explain the code base. Now they don't even need me to add kind of small features and to add improvements and to do things. So, so then you think, well, why can't I just hire anybody? Can I just take a guy off the street who doesn't know how to code, doesn't really understand the product, doesn't really understand anything, and maybe, just maybe, they're a little bit lazy. They're not willing to work hard, get their hands dirty. Um, could I just get somebody like that, throw them into my team, and hope that they make contributions? Well, let me give you one anecdote of why that's not the case. I hired someone uh recently and then unfortunately had to let them go and so this was transparent to them. It was kind of like a trial period. Uh we had a very short interview process and so the job was going to be the interview and they were working part-time so you know they still had their full-time job and this person had 2 years of experience and so they knew how to code. I mean this was not somebody off the street. This wasn't somebody who was just a student. This person did know how to code, but there was an issue. They made some pull requests and the poll requests were kind of working and it was all written with AI, which is fine. I encouraged it. Use AI as much as you can, can get away with it. But you have to know when you can't get away with it, right? You got to make that judgment call. Still, a person's got to do it. You can have all the AI code reviewer tools that you need, but even looking at certain companies, the CEO of one company is telling you, don't trust the AI review tool completely, at least not yet. So, this person had a lot of issues with their code. They unused or undeclared CSS variables, and I was just kind of confused by that. I was when I actually took a second to look at the code, I was confused like, what is going on with this code? You kind of just made a mess. And they had some merge conflicts. Uh, and so, uh, before I reviewed the poll request, they resolved them themselves and then they had some issues and then they used AI to kind of fix those issues. And they used AI to resolve the conflicts in the first place, which is kind of what made the mess. And so then, you know, they I asked them what happened, like what's going on. They said, you know, I figured it out. It's all good now. And, you know, when I actually did finally take sit down and look at that poll request, you know what I saw? I saw some pretty awful code. I saw large chunks of code that I wrote that were deleted. I'm asking the question, what's going on? Why is that all that deleted? And so then I asked the question like, did you know, did you read the code? No. Okay, so you're thinking like, you know, that's like, you know, I kind of confronted the person, right? Well, they actually didn't say no. They didn't say anything. So then I'm I didn't really know what to say at that point. And then so I was so I still gave them a chance. Like I just fired them on the spot or something. um you know but um yeah and then uh even after I asked them did they read the code they came back with a revision they said okay I fixed all those the issues that some of the issues that you pointed out everything's good now I go back and look at the code I'm like you must not have read it because all it takes is just to read the code and even try to understand it you'll see you just deleted chunks you don't know why you deleted them so clearly something is just missing and you guys aren't going to believe me and I'm not trying to be a jerk when I say this, but you know, hiring is really, really hard. You know, a lot of people think they deserve a job, and you know, maybe they do, but can you do you actually have the skills? Do you actually know what you're doing? And the good news is that a lot of those skills are the same that they were before this AI programming stuff was happening. And that's why you're seeing the interview

process. Even when they introduce AI programming, the concepts are still the same. You still have to read the code, understand the code, explain the code. When you implement a solution, you have to ensure that it is correct. Ensure the correctness of a solution. How do you do that? You usually with te with tests, but also by understanding like the problem you're trying to solve. Like typically even with DSA problems you have to understand the formal requirements sometimes the output format can be different very important to follow requirements that's always going to be the case in any kind of engineering so certain things are changing but certain things are not changing guys any time you write a piece of code whether it's AI or handwritten or not you know do you and I'm asking you this question for real. Do you have a mental checklist that you go through? Because if you don't, that's a problem because I'll tell you right now, any competent developer, and I'm using that word kind of strongly, right? You might get offended by it. I'm not trying to offend anybody, but in my view, any competent developer has a checklist. You know, at the very least, they ask like, is this solving the correct problem? That's the number one thing to ask. Are we solving the correct problem? Are we solving it correctly? Are there issues with this? Is there something I don't understand about the solution? Is there an edge case I'm missing? The single most important thing about any solution, whether you're talking about a DSA problem or a system architecture or a low-level di type thing, it's the single most important thing you got to ask like am I actually doing this correctly? Do I fully understand the question? And you know with all this use of AI and copy pasting you know back and forth and all that stuff some of these habits get lost and we are seeing it with a new generation of people who did not learn the fundamentals and did not learn these things. So one you solve for correctness. Uh number two uh performance and performance you know most of the time it doesn't really matter like when would performance matter like okay I have like a relatively small array when I say small it could be 10,000 it could be a thousand. Um maybe I'm doing a linear scan through that array when I could be using a hash set or hashmap. Well let me tell you something. Maybe in a DSA interview that'd be a big deal but typically in the real world it's not. I'm doing linear scans all over the place in my codebase and sometimes it's actually faster than using a hashmap uh for certain reasons and you know we don't need to get into cache locality and stuff like that hash collisions all that but those concepts are still they're still there we haven't fully abstracted all of this stuff and some of the other things that are important are like security now that's not usually going to come up in a um DSA interview but maybe in an architecture type interview security would be pretty important and that kind of falls maybe into correctness but another thing that you consider and when I say the word consider I don't necessarily mean you make it perfect but maintainability you know when you do a solution you have to understand the solution there's pros and cons to everything if you weren't taught to think in terms of pros and cons that's a problem if you weren't taught to think in terms of like we're trying to find the right solution for the particular problem not just you know just force a solution in there that's a problem because then how do you even know what the AI is giving you is good or bad if you can't tell me the pros and cons what's an alternative way to solve this problem and why this one is better than the other that's kind of a problem and maybe a lot of the times it really doesn't matter what the solution to a problem is. It's like I said, if you are scanning through trying to find an element in a thousand elements, uh a target element, use a hashmap, use an array. It really doesn't matter. But you should be able to say that that, okay, it doesn't matter because of this. You shouldn't, you know, what I'm saying is there's a skill that's been lost. It's called thinking. You have to think. You have to think yourself. Now maybe you don't have to think about certain implementation details like you used to. People say the thing, well we have AI now. Why do we need AI can solve every leak code problem, every algorithm problem? Why are we still testing DSA skills? Well, let me just tell you that we already had the solution to all these algorithms. The companies aren't hiring people to solve DSA questions. there was

something else in there. If you don't understand that, I'm just kind of perplexed. If you've actually thought about like why are companies doing it this way? You've thought about it for so long and you can't understand that they're not just trying to find people who are good at DSA so they can jump on the job and just solve a bunch of DSA questions. I think you're missing something. Um, and again, I'm not saying it's a good way of interviewing. Um, but I I'll be any day of the week I'll point out all the different flaws and the other types of interviews as well, even if there are better alternatives than DSA interviews. And of course, I know people are going to say I'm biased. Um, but a lot of those same people said that DSA was going to be dying in 2025. They said it 2024. 2023. And uh you know when I quit my job at Google uh three years ago, that was right after chat GPT came out. Everybody was saying DSA is dead. Coding is dead. Imagine if I had listened to that. Imagine if you know you were getting a CS degree or whatever and you said that well coding is dead. I should just give up. I should go do something else. You would have shot yourself in the foot. I would have shot myself in the foot. Neat code is just continuing to grow. And you know, maybe you'll die tomorrow. I'm not going to have a mental breakdown if that happens, but uh in terms of like what's going on and when you try to take a second to look at things objectively. Um it's very clear to me that certain things have not changed yet. It's hard to say what's going to change and when they're going to change, but obviously it's going to depend on your circumstances. If you are preparing for big tech interviews or if you're preparing for a company that you already know, well, just take a look at their interview process. At this point in time, what I'm seeing is that most companies are still asking some variation of DSA interviews. And if they are, you should prepare for DSA. Now, if they're not asking DSA or if you're interviewing for other companies, startups, companies that maybe have no sort of DSA types of questions, well, then don't prepare for it. Then don't. But what I'm seeing is that at the very least definitely at entry level and interns definitely we're still seeing a lot of DSA still seeing it at mid level senior and of course if you are preparing for DSA interviews I still do believe and plenty of other people believe that neat code is one of the best ways and honestly I think like at this point it's not even opinion anymore. I genuinely think it's the best way to prepare at this point. I can give you a dozen reasons why. I guess I'll go quickly. We have patternbased learning. It's simple. So you want to get started. You can go through one of the lists. Typically people go through neat code 150. Okay. You get stuck on a problem. Well, there's solutions for every single problem. Every single one. Look at it. So you got your video solution. Typically most of these are free. You get your written explanations. You get code in every language that you probably want to prepare in time complexity analysis. Everything. We now have a discussion system. You can solve the problem. The code execution is very fast. Let me show you how fast the Python code will run. Like just running a quick test case, it's like 200 milliseconds. So I'm going to click it right now. You might not even believe it's going to be finished. See, it's finished. I'm going to click it again. I'm going to keep clicking it. And yeah, you'll get rate limit exceeded because you can just keep going very quickly. And we added other features as well. So let's say take a look at this piece of code over here. Let's say you had a bug in there, but you don't really know what the bug is. So maybe it's an off by one error. Those can be pretty tricky to catch, but once you kind of know to look for them, then you start to look for them. So over here, if I change this code to I from I +1 to just I and then I submit the code, it passed the first test case, but then it didn't pass the next one. So it kind of might be hard to debug. Let's suggest a fix for it. And um what you'll see is that it gives a nice little uh UI. So you can kind of see everything in one place. I'll go ahead and move my camera. Uh and on the right side you can also see the explanation of it. So the original code had a bug. The inner loop was starting at index i, meaning i would equal j. And when that condition happens, it's always true because you're always comparing the same element. So what I like about this is that, you know, people are abusing AI. But that doesn't have to be the case. AI is this wonderful tool. It makes learning so much easier and better. And I think that this feature is already getting a ton of usage. people are finding it very helpful. I definitely would have found it helpful when I was preparing and so I hope you do as well. So in here you can see why the change was made. You can accept it or dismiss it or you can maybe reject everything and then just kind of fix it yourself so that you can kind of reinforce the learning. And maybe you don't want the fix. So let's reject that. You didn't want the fix in the first place. You're kind of self-motivated. You want to figure out that thing yourself. So let's ask for a hint. And sometimes you might need a hint when you have no idea how to solve

a problem. Sometimes you might need the hint when there's just a small mistake. So in here we will generate some hints for you based on your code. And if you uh are still not able to figure it out, you can just look at the solution. So open the solution. Um and maybe you don't need the hints, but you're kind of just stuck on the code. So you just need a couple code snippets like how do you use a hash set for example to uh possibly arrive at the solution. So, this will give you relevant code snippets and there's a ton more that I think I could add because, you know, when I'm a problem solver, I like to think deeply about problems. about my users. And I think I know most of you guys pretty well. You guys come to my site because you don't want to solve DSA problems. You want to spend the least amount of time solving DSA problems as you possibly can. So I have a feature that I'll be launching in a few days which is going to help you do exactly that. So for example if you were to tell me hey neat I have one month to prepare for my coding interview and so I have these time constraints I'm preparing for these companies and can you create a study plan for me? I'm going to say, okay, I'm going to give you enough problems uh that you can solve in that period of time, and I'm going to give you the ones that are relevant to the company that you're applying for. And third, I'm going to be intelligent about it. So, I'm going to say that, okay, we have all these patterns that could be asked by companies or that particular company, but DP is really hard and graphs are really hard, right? Advanced graphs are hard. DP is really hard, but let's say DP comes up one out of every hundred questions, but maybe advanced graphs comes up five out of every hundred questions. I'm going to tell you to study the advanced graphs. So, you know, when you actually sit down to think, you realize that there's so much to think about. There's so many choices to make. There's a lot of complexity in life, and it's always fun to find solutions to those problems, at least for me. And you know, whether we're coding for the rest of our lives or doing something else, I don't think is going to go away. And I love thinking and I'm going to keep doing a lot more of it. I'm going to become a better and better thinker. Thanks for watching and take