Engineering video
14:51 · 19 August 2025
In the speaker’s explanation, bogdan argues against abandoning software engineering because of AI or market fears. He points to engineers moving into other industries and distinguishes generating code from delivering production software.
Hear it at 0:00 →In the speaker’s explanation, aI tools frequently slow down production-ready software development because cleaning up generated code creates extra overhead, and major tech companies downsizing was largely standard corporate control rather than AI replacement.
Hear it at 1:45 →In the speaker’s explanation, modern AI firms are primarily machine learning companies that require traditional software engineers for distribution, APIs, UIs, and infrastructure, meaning mainstream developers possess broader job versatility than specialized ML engineers.
Hear it at 3:41 →In the speaker’s explanation, system design and architecture cannot replace coding skills because great architects are fundamentally expert coders who understand the daily difficulties developers face when extending maintainable systems.
Hear it at 6:15 →In the speaker’s explanation, quitting a day job to vibe-code a SaaS product ignores the reality that business success relies far more on marketing, operations, and finance than on coding, making employment much more reliable.
Hear it at 8:07 →In the speaker’s explanation, switching to niche programming languages for perceived safety against AI provides fewer job openings and limits upward mobility into staff or principal roles compared to mainstream languages like JavaScript.
Hear it at 9:34 →In the speaker’s explanation, transitioning into product management under the assumption that AI will write all code turns engineers into low-value generalists, ignoring that product management requires distinct years of experience across pricing, scope, and stakeholders.
Hear it at 11:39 →In the speaker’s explanation, relying on side projects can make candidates appear like junior hobbyists, while open-source maintenance is often poorly paid and resembles unpaid volunteering.
Hear it at 12:47 →So, let's talk about the worst career advice we see for software developers in 2025. The number one bad advice is that you should just quit your software engineering career. We're seeing a new trend on YouTube where people are quitting their software engineering jobs even if they pay very well. And it's usually the same people that were doing the day in the life videos 3 years ago where they were bragging about doing pilates at work, coding for like half an hour, having an overpriced latte, and making like half a million as a developer. And so, why is this happening? And why is this a very bad
idea? Well, before I tell you that, let me introduce myself. I'm Bogdan. 5 years ago, I started a senior dev with my brothers. We're both developers. And over the last 5 years, we helped 350 people get to senior developer and beyond by helping them, making sense of all this and ignoring all this bad advice. Now, why are people telling you to quit software engineering? Well, because AI will basically replace software engineers fully or partially.
It will automate the job. It can generate code much faster than a human developer and we are all doomed writing code it's useless nobody will write code now that's very false we've seen recent studies where AI productivity increases are far less than some Alman and theropic CEO and so on they want you to believe when it comes to production ready software AI does usually more harm than good and they're also saying that layoffs have pushed too many people in the market so it's kind of a combination of its factors but the reality is that a lot of The engineers laid off in 2023 and 2024 actually found a job in a
nonpure tech companies. So they go to other industries. Those industries needed software engineers but big tech was basically monopolizing all the talent for the last couple of decades because they have access to very cheap capital and a lot of monopoly businesses. Now at the beginning of this year Zuckerberg said AI will replace all mid-level engineers very soon. The Microsoft CEO said 20 to 30% of the code at Microsoft is now written by AI. But we are seeing new evidence and studies on real life software engineering jobs where they realize that actually AI can even slow you down because yes you do
get a lot of upfront code generated but cleaning that up making that workable it's actually slowing people down. There's a lot of evidence on that. Feel free to check our channel. We have a lot of videos where we dive into coding with AI and what is the reality behind it. The other reason they are telling you to quit software engineers is because there's too many CS graduates. But we've seen CS enrollments are dropping even in the most prestigious colleges and universities. So we predict that in two to three years you'll have again a shortage of great engineers in the sector. And the other reason is people
are claiming that the only reason why we had software engineering jobs is because we had low interest rates. Yes, that's true. Having low interest rates created a sort of a bubble in all jobs, but the economy is cyclical. And so we are now in a downturn, but then the upturn will go again and then you'll have again shortages. So people will look for software engineers just like they did before, especially very qualified software engineers. Even if a lot of people are trying to fearonger about the layoffs, we see that the actual people employed in the tech sector went down only by 1.7% since the peak. And so yes, maybe the
layoffs looked like very big numbers, but you have to take into account that tech it's a huge sector. So even if they lay off 1% of people, then it still feels like a lot of people. And also take into account the fact that when big tech was laying off people, it was also laying off marketing and HR. It's not like they just laid off their engineering force. So it's definitely not an AI. it was rather a control downsizing and they will get back to hiring when they will realize that AI it's actually a net loss in their balance sheet and that will happen sooner or later. Now the second advice is to learn AI because if not it will
replace you or someone that knows AI will replace you. So go ahead start learning Python, start learning machine learning, start building your own LLMs. The reality is there's a lot of people making a lot of money maintaining legacy code. And I really like this memory because he's basically cleaning his tears with a lot of dollars. And that's how experienced developers feel when people are saying, "Oh, AI will take your job." You be like, "Sure, yeah, as long as I'm making all this cash." So number one, AI is very far away from replacing software engineers. If you are shipping production ready software, you
already know that. The second is modern companies like OpenAI, Antropic, they are not AI companies, they are machine learning companies. So they build this huge machine learning models, then they wrap them around with some software. They duct tape them into looking agentic or whatever the trend is, and then they ship it and claim that's AI, but in reality, you just have this LLM, this huge large language model that they kind of play with it and productize. Now most machine learning products, they need a lot more software engineers to make them reality than they need machine learning engineers. You have a lot of
distribution, you need to build APIs, you need to deploy, you need to manage infrastructure, you need to build SDKs, you need to build UIs, you need to manage traffic. All those things are done by developers. And that's why you have so many developers working nowadays where computer science in software engineering it's the biggest engineering discipline in the US because business runs on software and you need all these people to make everything reality to make you know banking reality to make healthcare a reality. You need those information systems. And so yes, you can go ahead and study machine learning, but
that's like saying that everybody in the car industry should become like a engineer. You'll always have an engineer because the car needs an engine, but there's a lot more that you need in the car. There's a lot of manufacturing happening to make that car a reality. And if you're an expert in manufacturing, why would you go ahead and throw everything away and go be an engineer because they told you so? And so it's the same with software. people building the model. You'll have those people, but there's usually about a handful of researchers that are doing that. And we actually seen in 2023 and 2024 that the unemployment for data
scientists and machine learning engineers was even higher than for traditional developers because guess what? A traditional developer, a Java dev, a JavaScript developer, they can go work for a very small company and still help them by using, you know, low code and setting up their website. you can do a lot of things there and add value. Whereas someone that knows machine learning, they can only work for a handful of companies because they're very specialized. Number three is to forget coding and learn system design and architecture instead. Coding doesn't matter because AI can code. You should
look into system design. I think Satya Nadella said that a few days ago. Now remember, a great software architect is first and foremost a great coder. Architecture, system design, they are an extension of your coding skills, not a replacement. And so saying that you will be a good architect or good at system design without knowing how to code is like saying you will write a novel but you cannot write a paragraph. It makes totally nonsense. The main user of your software architecture that you're building or your system design. It's usually first the developers. That's why we build microservices. That's why we
build event driven because it's the developer that's going to use the system to extend the system. And that's why software architects are usually very good at coding and they understand all the difficulties a developer will face extending the system and then they rearchitect the system to accommodate that. The main reason why we have microservices, microrends is because we consider the developers extending the system as a user. Now how are you going to build an extendable maintainable scalable system if you're not a developer? You don't know what the developer needs. You don't know what the work actually needs. You don't know
where the work getting done. So again, saying that you'll go into software architecture or system design without excelling at coding. It's a very big fallacy. There's no difference between a great coder and a great architect is just a natural evolution. And so that is what a lot of people are doing. They add all the sophistications and agents and trying to get the AI to do something that you could easily do just by typing out the solution if you're a developer.
Now, if you want to know exactly how to become a better engineer and grow into that architect down the road, make sure you take our free technical assessment. We build it with interview questions and it'll give you a top level view of your skills. What are your biggest strengths and what are your biggest weaknesses and what you should focus on. Number four, build a SAS. Quit your day job. Build your dream. Forget the 9 to5. Just vibe code and you'll make a lot of money.
There's a lot of influencers that also jumped onto this train where they're selling you that you'll be on the beach coding your SAS and making a lot more money than you would on your day job and you'll also have magically a six-pack. The reality is if you can't get a job, you probably won't be able to build a successful SAS business. It's just a lot more difficult to build a SAS business than to sell your employment to someone else. And that's okay. Why would you do something in the most difficult way when there's an easier way? The skills required to build a successful SAS are not coding. So coding is usually the
last part. Usually you need someone that's really good in marketing. They have a great following and they can place a product. They can make sure that you at least get some users so you can get feedback and make it better. You have operations. You have the finance part. It's not just coding. It has never been. That's why successful software companies, they usually have a combination of a great CTO but also a great CMO, a great product person and access to funding. And so we do know a lot of people, a lot of developers that fell for this trap and then they come back after 2 years. They depleted all
their savings, the business didn't go well and now they are back to getting better at JavaScript and passing that interview and getting that senior job because they need cash. And what they realize is that no matter how hard the market is right now and how overcrowded, it's way easier to get a great developer job than it is to starting your own successful SAS business. Number five, learn a niche programming language because AI is not very good in those languages and it will not get place today. Well, the reality is that niche languages have niche opportunities. So, it's not like you have thousands of
cobalt jobs. It's actually very few. So, yes, maybe you make a bit more money on the short term, but in the long run, if you're unemployed for let's say 6 months, you've lost all that benefits. Whereas if you know a mainstream language like JavaScript or TypeScript, sure maybe you're not able to demand like such a big compensation as someone that's more specialized, but you have the job security because there's always someone that needs somebody doing TypeScript or JavaScript. Remember that if it's written in a computer, you can write something on paper, then an LM can learn it. So the reason why we don't
have good LLMs for cobalt or Haskell or any of those languages is because they weren't trained on that data set. But as soon as we get that data set, as soon as they get into private code bases and they actually learn, you'll have specialized LLMs for cobalt and for all these niche languages and they'll do a lot of work there. It's not like you can gate keep the LLMs from learning it. And so you are basically changing your career to go into this imaginary field that people told you that you make a lot of money, but you don't know it. It's not sure. And it's also there's a lot of people there that have more experience
than you. You want to really listen to the market. And if people paid you for JavaScript, TypeScript before and they pay very good money, you keep doing that and you keep getting better. Also remember if you are doing a more mainstream language like TypeScript and JavaScript, it's so much easier to go upwards. It's so much easier to end up being a tech lead or a principal or a staff engineer because those languages have big code bases that they need to maintain and big companies running that code. Whereas in Cobalt, you probably, you know, at most be a senior engineer.
It's not like you can go and be a staff engineer just by knowing that language. And also remember that the biggest problem in this right now it's a lot of developers don't have strong fundamentals. It's not that they don't know enough languages but they don't know a language well enough to be really experts there. Become an expert at one thing first before you learn five new ones. And yeah unless you want to become one of those folks then you don't need to learn cobble at least for now. Number six bad advice is learn product management because you'll be a product manager for the will do the coding the
LMS will do the coding and you'll just kind of manage them. This is a very bad idea because you will become a low value generalist. You will know everything and know nothing too well. Let's be honest. Everybody can query GPT. Everybody can install cloud code. Everybody can add hooks to it. Those things are not hard. And remember that AI it's a much better generalist than it is a specialist. But it really sucks a task where you need a lot of depth and precision. It really goes in cles. It doesn't have enough data sets to really make a decision. And that's where an expert will excel.
Product management done well. It's a set of very different things. stakeholder management, scope, offer pricing, and it's just another career altogether. So to get really good there, you need years of experience. If you ever worked with a great product manager, you'll know what I mean. And so if you're a developer and you got into this career path is because of something. Is because there's something in software dev that really attracts you is because you're probably good at it and you want to stick with it until you become an expert. Don't go in and follow those people and then try five different things only to land back
in software development. And advice number seven, it's do the same old things that never worked. Like do side projects, do open source, start your own YouTube channel, become an influencer. And so people are saying, "Oh, do a side project." But in reality, side projects will make you look like a junior hobbyist. If you go to a senior interview and you tell them you did this side project, they'll already look at you like a mid-level junior. And most of times disqualify you even if they don't tell you. So, you know, they might be like, "Oh, this is so cool. You did this a cool, but you're already, you know, in
the happy meal category. You're already a junior." The next advice is to do open source which is very hard to do well. It's very low paid and I know a lot of open- source devs that are very poorly paid. They can barely pay bills and then they end up maintaining the open source project while they have a job. So even if you're crazy good at it, it's almost like volunteering. I mean I know people do it for a calling, but I also know people have a mortgage. They need to pay bills. And so we got to be pragmatic here. And the number three is create content. Yeah. Go online and do your YouTube channel and have your own thing.
A lot of the people that went into that direction are now looking for a developer advocate kind of position. It's a different job. You're just creating content and basically getting developers to use some sort of a product. I think it's great for certain personalities, but if you like building and you like building software, you'll enjoy that more than just doing YouTube videos. It might seem very exciting at first and you get a lot of validation because you're in front of the camera, you get a bunch of likes, but it's a job, okay? Okay. And at some point it's going to get boring and doing the next
video is just going to be as exciting as reviewing the next pull request. And then it's crypto service equity and whatnot. So there's going to be thousands of distractions and shiny objects. People will try to steer you away from what you're doing. But in reality is if you stick with it, if you get good at it, even when it gets boring, even if when it gets tough, it will pay back and you'll end up having a great career. You'll end up having a lot of options and you'll end up making a lot of money. Don't get distracted. Get better. Now, if you are a JavaScript developer, we help you get to senior or
you don't pay. We help people go from mid-level junior to a senior full stack on the JavaScript pad. We've been doing that in the last 5 years with over 320 people. Check the links in the descriptions.
