Imagine this. Two engineers interview

for the same role and both use AI coding

tools every single day. Now, one of the

engineers gets the job while the other

cannot even get past the system design

interview. How do you think two

engineers can have such different

outcomes if they both use the same 10x

AI tools? Well, having grown from junior

to senior myself and interviewing

candidates at tech companies, I know

what the difference is between a bad and

a good candidate. And if you are a

junior right now, you are probably

getting distracted from learning what

really matters. Half of your feed says

AI is going to replace you. The other

half says that learning to code is dead

and that you can just vibe code

everything. But what do you [snorts]

actually believe about all of this? Let

me share you a story that will help you

form your own opinion instead of just

listening and following others. A hiring

manager recently told me about two

candidates he interviewed backtoback for

the same position. Let's call them Fu

and Bar. And so Fu walked through his

projects and could explain his decisions

in every single interview. When they got

to the system interview round, he talked

through trade-offs, why he might go for

a NoSQL database, why he'd structure the

API in a certain way, and when the

interviewer pushed back, he could defend

his choices or acknowledge when a

different approach would make more

sense. Now, the second developer, Bard,

had an impressive portfolio, a bunch of

AI assisted projects with polished

demos. But when they hit the system

design round, it fell apart. Why would

you use a SQL database here? Could be

one of the simple questions, and he

wouldn't know. What happens when this

service gets 10 times the traffic? He

couldn't reason through it and explain

how he would scale up the service. He

didn't even have an explanation of why

he used GPT4 for one of his AI projects,

which is already a deprecated model at

this point. Good technical interviewers

see right through this. We know that

you're using cloud code to generate this

code and that it picked GBT4 from its

training data as the most recent model,

not a conscious choice that you made as

a trade-off. And that is not really a

problem as long as you can justify the

code that is generated yourself in an

interview. And Bar in this case built

things without understanding why it

worked. So Bar could ship features all

day long, but he could never explain the

decisions behind them, which meant that

Fu got the job. Because we as technical

interviewers may not be able to see the

difference in AI generated code, but we

will immediately see through you once we

have a live conversation about software

for half an hour. In this case, FU can

be trusted to make decisions when

nobody's looking over their shoulder,

while bar cannot be trusted. You have to

understand that companies don't care

about whether you can oneshot a to-do

app. They care about whether you can be

called when their payment system goes

down and nobody knows how to fix it.

They want to know whether they can hand

you a multi-million dollar project and

trust you to make the right architecture

decisions and be responsible for them.

And being able to make those judgments

is what separates a junior who gets

stuck in hiring with a junior who does

get hired and even gets promoted

quickly. So ask yourself, if someone

asked you a certain decision that you

made in your last project, could you

explain it? Because when requirements

change or something breaks, the person

who understands the system is the person

who can fix it. Poenl, the math Olympiad

coach, put this perfectly. He said that

using AI to do your homework is like

driving your car one mile for exercise.

So think about that, right? You're not

really saving time because you're

skipping the entire workout. And the

workout here is the point. When you

struggle through a bug for two entire

hours, you are not wasting time compared

to letting cloud code do everything

because you are building a mental model

of how the system works. That mental

model is what lets you debug the next

problem in 10 minutes instead of 2 hours

and actually get through technical

interview rounds. So don't get me wrong,

right? AI tools are incredible. I use

them constantly. But there's using AI to

learn faster and then there's using AI

to just skip learning entirely. The

junior who uses AI to explain concepts

to suggest approaches that they can

think through themselves, well, that

person is truly learning way faster than

anyone who does not use AI at all. But

the junior who just copies AI output

without understanding anything. They are

building a house of cards that will

collapse the moment they try to get a

truly technical role. Which one are you

right now? You are either scared of AI

because you are on social media too much

or because you are not such an engineer

yet who can handle the real technical

interviews. And there is still a

difference between a developer and an

engineer. A developer just writes code.

An engineer understands complex end

to-end systems. Developers could be

replaced by better tools, but engineers

cannot because engineers are the ones

who decide what to build and not just

how to build it. They are the ones who

catch when the AI suggestion would break

something downstream and can take

responsibility for it. They are the ones

who can debug a production issue at 2

a.m. without copy pasting error messages

into chat GPT and hoping for the best.

No matter what you might read on social

media, there is no perfect self-healing

agent used in real production scenarios

because companies will always need

people who can think through problems,

weigh tradeoffs, and make decisions they

can defend. That's not going away. no

matter what people might tell you on a

platform like X. If anything, it's

becoming more valuable as AI makes the

code itself cheaper. So, if you are

worried about your career as a junior,

this is actually good news. The path

forward for you isn't to try and out

code AI. I write most of my code with AI

nowadays, but you have to become the

person who knows when AI is wrong. And

so, if anything, here is what I want you

to take away from this video. Don't let

social media decide your career for you,

including this video. Have the people

posting about AI have an agenda. They

want clicks and they want you to feel

fear and they want you to especially

feel like the sky is falling, right? But

if you actually talk to people building

real projects, hiring real people, and

shipping real code, you will hear a

different story. I can't believe I have

to say something so obvious, but

engineers are needed everywhere.

Healthcare, finance, logistics, energy,

every industry is still requiring a lot

of well-maintained software, and they

all need people who can think, not just

prompt. Sure, maybe there will be less

engineers in the future, but if you're

watching this video right now, then

you're understanding already the

mentality that you need to become future

proof. And the fun part of building

things is having your own contribution,

your own twists and ideas that you bring

to the table. And that doesn't go away

just because AI can generate, you know,

500 lines of Python in 2 minutes. If

anything, AI gives you more leverage to

bring your ideas to life faster. And you

probably have better ideas than the

standard to-do list app that people are

generating every day with AI or the 1

millionth Tetris clone, right? I believe

you have much more capability than that.

So don't be afraid of this moment where

everyone is doom posting about AI. Be

excited because if you take the right

approach here, AI will become the

biggest career accelerator that you've

ever had. You just have to put in the

work to truly understand what you're

building and stop scrolling so much. Put

the fundamentals first and then put the

AI tools on top. The engineers who are

thriving right now learned how these

systems work first. Why do you think

that senior engineers are just able to

make so much progress and why there

seems to be so much doubt about the job

security of juniors especially? Well,

it's because they already have the

fundamentals and then they can add AI on

top of that to truly move faster.

Seniors can review AI generated code

because they know what good code looks

like. They can debug AI suggestions

because they understand the underlying

logic. So if you do things right, you

are not competing with AI because you're

building the skill to direct it, to

orchestrate it. And that skill still

requires software understanding. Now, I

have an agenda, too, because I have a

lot of free resources to help you with

your engineering career. And you can

start with the free AI engineer starter

kit in the description or decide to

watch the next video and keep

distracting