Видео

can you please provide the ppt link?

Really nice video , It helps to refresh interview

What do you think of the points in my video? Leave a comment and discuss! - Increasing trend to make backend developers do DevOps work. - Increasing trend for DevOps Engineers to be developers. - Why would a developer even want to do DevOps? - Ever increasing toolset (and studying) and every employer expects you to know their exact stack. - Knowing an equivalent stack is respected less and less. - DevOps in a startup - too much to learn and basically the company's dog. - Platform Engineering and real SRE the only light at the end of the tunnel. - Everyone else doing "Glue Engineering" - highly unstable and unclear future, even before AI! - For "simpler" Cloud Infra engineers, cloud migrations are drying up. - What to pivot to? Without FAANG style savings.

I'd love to hear your personal Terraform tips and tricks, please leave a comment and share with the community!

Thank you for the great video can your make one on how to make a good looking GitHub and read me for documentation

Can you make a video on how to document and your projects on GitHub and setting up a resume and portfolio website

Ok so did you actually create your ball program with chatgpt or not? Have you actually tried it and it worked?

Simple, ask the manager why they think its okay to put our IP onto a random server to be analyzed and potentially stored for the AI to use as data in the future? Listening to this guy saying "I dont care about syntax, only logic" shows inexperience. The structure and syntax of your program is just as important as your logic. However, syntax issues have been solved without AI for decades with an IDE. Structure and logic is the important part. Writing well made programs with an architecture that is maintainable and flexible. Logic so that it meets business rules. Etc. AI can basically help you with snippets here but it cant really produce anything over an existing code base.

I work at a small company an thankfully my boss gets it - I use AI but it's primarily to get information. I use documentation as well, but I've found sometimes AI is faster and easier to find what I'm looking for than searching the web. It's really helped me learn things and use new languages too, if you are smart about how and when you use it it's genuinely a great tool.

Me: Ez way to explain.i am human that’s ai of course it’s faster 🙄

Let the manager do it themself with AI and deploy it to production and see if it works The question won't come up again after that.

We need to know what kind of task that was given also compare both solutions. we can not just way the manager did wrong.

99% of all companies are small enough to only need one disciplined developer. Dev ops is really nothing more than a useless admin that adds time to a project introducing documentation and communication, taking time away from actual development, and this video perfectly illustrates it. LLM's do produce readable code - that is not an issue. As for testing, and you mentioned never been acceptance tested...how could it be? That by definition happens when you turn the project over to the client. As for having policies on how to structure your code...is about as useful as a teapot made out of chocolate. If you show me someone who insists that code that actually performs well should be re-written to have a { on a new line rather than after the ), then I will show you someone who should receive a high velocity impact to his face. What LLM's actually gets wrong is efficiency, memory usage, knowledge about your hardware and software infrastructure, equations and what to use in different scenarios. It may even reply with things that are deprecated and does not work anymore.

The whole premise is off to be honest. Taking 5 minutes to generate some code is incredibly slow for an ai. So if somebody says that they are obviously including prompt writing time into that. And to that manager I would be suddenly excited and ask them how they wrote the perfect prompt in such a short time. They got production ready code out of it? With all the supporting features, and it didnt get confused and make any bugs or miss features or implement any details against your intentions that you inferred in your prompt? And you wrote that in under 5 minutes? Amazing! Lets ship it then, and you should be exploring your new found super skill. Developers are going to be so thrilled to learn how you did it! We are going to be rich!

I will never code without ai again but in its present state it creates some really awful code. I am a different coder now. I mark homework produced by the ai. It lets me tackle some bigger challenges than before, where i would have got stuck and reduced the scope working by myself. So it does feel like its better but it also adds bugs, gets things wrong, uses bad practices, regresses and loses features already implemented. At the moment i would say if you are not a professional programmer be very wary of ai code. It will confidently challenge you to find its mistakes while swearing blind it has made none.

I tell you if the layoffs in tech industry are happening fue to these managers reporting that they don't need devs coz ChatGPT can do the same stuff they are going to suffer a lot

It is very hard to explain to non-coders why AI submitted code isn't truly usable. To them if the outputs look even remotely similar to what an experienced programmer might produce, they will inevitably think any explanations are just excuses for incompetency. I think the most important thing to stress in such confrontation is that human beings can actually think about the problem space and work independently from the inputs. The second point to stress, is that the AI guesses a solution, and while that might work for quite a while, once it guesses wrong, the non-coder manager won't even be able to tell what went wrong and they'll need me to fix that mess. Software is more than just the text files.

tell them to just use AI then. maybe it is so easy, maybe it's not let the objective reality decide. i mean, it's not realistic, either the programmer is exceptionally bad or manager simply lied, which is more probable.

If a manager demands to know why it took me 3 days instead of 5 minutes I would simply reply with a, "It is a little thing called 'Quality Control' not sure if you know anything about that."

But how could you meassure the efficiency of a developer? In the video you say that you use it to not worry about syntax. Well, that's what LSPs where made for (autocompletion and syntax check) Me personally don't use AI for coding BUT I used them for boilerplate, formatting and all of these boring things that LLMs are really good at. On the other hand, you say that AI speeds up development time. How do you meassure it? Most of the benchmarks about AI speeding up developing time are also the same companies selling you the AIs (for instance, github claims that AI speeds up to 55% development speed while selling you Copilot) Otrer thing is that when you prompt engineer you seem to be VERY specific about the implementation (you ask him to use specific libraries, specific syntax, and break down the code to it's atomic functions to not overload the AI). Is it THAT different than implementing it yourself? After all, you gotta make the very specific prompt, you gotta check the code, you gotta paste and test if the code works, and if it doesn't for some reason fix it (or ask chatgpt to fix it! i dunno) and at this point I don't know if you wern't just better off coding it yourself. I'm not blaming or anything, I think the video has a lot of good in it. I agree in the fact that, AI or not, you MUST learn what you are doing and. I don't recall if you said it, but it seems that you don't buy that "AI is gonna replace us" bullshit and I really found interesting that you proxied some programming techniques thanks to AI. I liked the video, thumbs up and waiting a response. Cheers!😊

Why no link with the commands/presentation ?

Good video. I'm working on becoming a CE first, and then eventually, I want to work my way up to DevOps or SRE.

Thanks for the great videos. Can you make some videos on a great devops resume, online portfolio and even how to write a great read me for projects

*side-steps it and does the actual work*

Tool proliferation across the entire web stack seems to be by force of nature. There's some sort of problem. At least five different solutions come about two of them die off, but they're never really dead because somebody used them somewhere. The other three go on to varying levels of success. The primary vendor that caused the problem solves 30% of it. Core audience says the problem's not an issue anymore. People cling to their hard-earned solutions. Get a little bit better. Some of them die off. Eventually the primary vendor says oh look we've discovered sliced bread and they solve the whole problem in a completely different way that has its own new problems.

Great videos thank for your knowledge and feed back. Can you make a video explaining in detail how to emulate your coding setup

Interesting one, so with your experience in this data, I wanted to get in the tech industry through cloud/devops and move to middle east but does not look great. What would be interesting is to see in some months. Where are you based ? Did not find out the answer on your channel.