How to Stay Relevant In The Age of AI coding

Disclaimer: No AI was used in writing or editing any of this email… Why? One, because I love to write. It makes me think deeply about stuff. Two, because I hate AI slop. It’s everywhere. And it’s disgusting. Three, because if you give me your human time by reading this… I will at least respect it, and give you my human time writing it. I only use a grammar checker with it's AI crap features turned off. It only checks grammar mistakes, and it doesn't refactor my writing! Now, let’s get going.

Dear reader,

If you are a developer writing code for a living… You are probably worried about AI taking your developer job.

Stop worrying. There are plenty of ways to win.

Here are 2 strategies to help you stay relevant for decades to come:

1. Be the “human insurance” against AI slop

Sure, AI can code.

But it makes mistakes. Tons of them.

It hallucinates... It gets side-tracked... It generates random test cases... And it comes up with libraries it doesn't need.

Meaning... You still need a human developer to verify all that AI generated stuff!

(Or, like some “vibe coders” out there you can just push AI generated code to production and pray for the best… If you work at one of these companies, do yourself a favor and start looking for another job right now… You will have to do it either way cuz’ these guys will soon go bankrupt!)

The ability to distinguish between clean code and AI slop is becoming the most valuable skill a developer can master these days.

So, position yourself as the developer who can fix things when AI code breaks… The developer other people go to, when they run out of tokens. The engineer who can get stuff done even if Claude Code is offline.

The one who understands what's actually being shipped to production.

Become best "code verifier" in your team and you will have a job for decades to come.‍

Our Software Mastery Program was designed to help you become that well-rounded Senior Engineer, who can provide tons of value in the age of AI. See if you are a fit here.

Here’s a little roadmap to become the best “code verifier” in your team:

a) Master the fundamentals.

AI is the code monkey now… You have to become the real Senior Engineer.

If you know exaclty what's going on under the hood… You will quickly figure out why the AI generated code is breaking… Regardless of what your AI chatbot is saying.

Be warned: You will need to have the technical depth most developers don’t even bother to reach. That takes a bit of work.

For example, many Senior JavaScript Engineers still don’t know how the browser actually renders stuff on the screen… Or how it fetches data…

Or how CSS actually works, and why.

The devil is in the details.

b) Master debugging.

I don’t mean “cosole.log()” until you find the problem… I mean serious Senior-level debugging.

You probably don’t like debugging. I know I didn’t.

It was the thing I avoided the most in my Junior years… It was also the most valuable!

That’s how the Senior devs in my team saved the day… They knew how to identify problems that nobody could… After that, fixing them was the easy part.

Well, with less experienced engineers using overusing AI… Debugging will become the most important skill you can bring to the table.

So, spend as much time playing with debugging the tools you use everyday (your IDE, the browser, etc).

c) Master the code review.

Start by reading tons of code.

Then, develop a system to make quick high quality code reviews. For example:

1. Big-picture scan to see if it integrates well with the codebase

2. Overall logic and functionality testing

3. Syntax and clean code check

(ohhh and bookmark this email for your next technical interview... they might as well ask you "how do you review your AI generated code...")

By the way, if you want Bogdan and I to go deeper into this, let me know. We might make a full video about it.

Senior Dev Tip: A lot of the developers think that the main thing they get paid for is coding… And I don’t blame them. Most of the “developer productivity” talk is obsessing over on lines of code.

In reality, coding is only a tiny fraction of the value a developer delivers. With AI writing so much code, that part got even tinier.

So, what part of a developer’s job is more important than coding?

Well…

What do firefighters get paid for? Or notaries? Or nuclear plan operators?

Many professions in the real world don’t get paid for direct output. They get paid for “just being there in case the sh**t hits the fan”!

Soon, developers will get paid more for overseeing AI code… Than for writing code themselves.

This is a very important job.

The biggest challenge of our times is that AI cannot take responsibility for the code it writes.

So, a human developer will still have to “certify” the code is clean, and can be shipped to production.

This won't change because LLMs are by default stochiocastic and probabilistic.

Hallucinations are a feature, not a bug. What makest the model creative... Also makes it more prone to errors (lmk if you want me to write more about this).

So, become that developer… The one who can “certify” things… And you will never run out of work!

Our Software Mastery Program was designed to help you become a well rounded Senior Engineer, who can provide even more value in the age of AI. See if you are a fit here.

2. Be the “bridge towards the agentic future”

Everyone from your CEO, to your CTO and VPs are probably sh***ing their pants right now… Their competition is already implementing AI.

Worse, if they don’t use AI, their investors will stop giving them money… And start giving it to businesses who do use it.

Disclaimer: I believe a lot of investors are behaving irrationally, funding worthless companies just because they “use AI to do [X]”. Is stupid, is a bubble, and it will pop. But that’s for another article.

Regardless, most of those business people don’t have a clue about how AI actually works.

Or how to effectively implement it in their company.

That’s your advantage.

You want to become the most knowledgeable developer in your team regarding AI coding and LLMs.

Learn about "model knowledge" and "working memory". About context engineering, RAG, MCP and agent harness.

If you are brave enough, deep dive into how these coding models are built. From pre-training, to post-training and RLHF (reinforcement learning from human feedback).

By doing that, you will become a developer who can say…

“Hey, this AI thing is happening, and I have the skills to help you prepare your codebase and your tooling for this transition.”

Once again… Become the developer who can help them make this “agentic future” a reality and you won’t ever run out of work!

(even if you don't fully belive all the crap Open AI and Claude CEOs are saying, this is what companies are buying right now... so sell them what they want... and give them what they need -- which is probably simple, uncomplicated code that works well, boring, I know)

By the way… this is exactly what Bogdan and I can help you with in our Software Mastery program.

Check if you are a fit… And book a call with me here!

Let's kick ass in the age of AI.

Take care,

Dragos