Well hello there so i'm currently in korea enjoying my

recharge and recently there has been a bunch of

talk about ai self-improving and just a lot of kind of news

around how ai is essentially replacing people's jobs and

looking around in korea there's so many young people and

like busy people just running around and it got me kind of

thinking about myself back when i graduated in like 2010

or something it was right after the housing crisis so i

remember finding jobs were extremely difficult now

obviously i'm not saying that the current market or

anything like that is anything related to that i don't

know actually

what was worse i guess technically when

i graduated it was officially a recession

um

but to be honest the current like data all of.

The data that we see, a lot of like the unemployment

numbers and stuff seems a little misleading for

some reason. Like this time feels

very different. Like this time because of AI and what it

could mean.

It just feels different than any other recessions

or downturns. Like it almost feels

existential. And I think a lot of comments that I get, a lot

of people that reach out, especially young people in their

twenties and thirties, they have a lot of concerns,

you know, what should I study? What should I do?

So in this video, I thought I would give 50 or so tips at a

high level

on what I would do if I was in my twenties again,

or early thirties. And I didn't already have like a

major career that I could like lean off of,

right?

Because the numbers clearly showed that the

jobs that are getting automated right now, Right now,

it's more focused to tasks that, like, junior engineers

or junior, like, tech workers would do.

And, you know, as these, like, as Cloud Code, Codex,

and these, like, agentic tools gets better and better, I think

a lot of CEOs and a lot of just, like, people who are not

actually doing the work

will go in and be a little bit too aggressive and, like,

replacing a bunch of people that they still need. Still need.

But at some point, there may be some reality that a lot of

the jobs that we used to do or we needed people to do,

AI may take over those jobs.

So this is not a doom and gloom video. This is more about

how can we adapt

and

kind of put our best foot forward.

This is more about like high-level thoughts that I have on

what I would do if I was in my like early 20s or 30s and I

was 30s and I was just starting out again because I

actually switched my career at 28

to software engineering so I was right so this is really

just advice

that I would give to myself back in the day. I'm going to be

teleporting a bunch of places as I'm making this while I'm

kind of exploring Korea

so I hope you guys enjoy some of the scenery and also

some of the tips all right so the first tip to level set is that

your feed is not exactly the marketplace.

The thing about your feed and like news in general is that

negative sentiment always sells. There's a lot of study that

says negative

posts, you know, rage baiting essentially gets a lot

more attraction. So all of the doom and gloom about

people losing their jobs and like doing a hundred posts,

a hundred applications and not and not getting anything i

think there is definitely validity to that but essentially

all of those posts are people who are still looking who are

still angry and who are still kind of deep in that trenches

so one thing you should really think about is instead of

just relying on doom scrolling you should really do some

deep dives onto kind of what the actual

data is around unemployment and what kind of jobs are

like not being posted, what kind of companies are hiring

and not hiring. I think like doing your own analysis and

deep dives and rather than just letting something spoon

fed to you, I think that is like a first step. So your feed

is definitely not the actual marketplace. So don't get

distracted by that. All right. So the next tip is around

never asking a barber for a haircut. Now, what this

basically means is that all of the AI companies.

AI companies essentially has a reason to hype up

the progress whether that is AGI or AI that like

self-iterates or clock code that just builds all the features I

don't code anymore all these kind of things there's

probably a lot of truth to these statements

but at the same time all of these companies have major

incentive to make sure that essentially the gravy chain

continues to go so making sure that people are kind of

bought into it people are invested to it people people

somewhat has a fear of missing out. Now, with that said,

I don't think this is all hype.

I genuinely believe right now that learning agent encoding

tools like ClockCode or Codex is probably the best thing

you can do for your career. And AI is definitely here

to stay. But at the same time, You should always take some

of these like predictions and all this kind of hype

around AI

in

with a grain of salt and really do your own research and

get your own feel

of how these tools are evolving. Now, the next tip is that

demos lies and productions tells the truth. So if you're

worried that essentially AI is taking over everything

right now, there has been real studies by MIT that said 95%

of AI adoption from major like Fortune 500 companies.

Companies had so far failed, and they have not seen like

real meaningful progress. Now, the study is a little old.

It's before the November kind of Opus 4.5 release. So

probably a lot of these things have changed by now.

But when I look at the landscape overall, and I do a bunch

of training on Cloud Code and Codex and these kind of

AI tooling,

I will say that most people are still are still not very sure on

how to leverage these tools

to their best abilities. And I think it's only like a few handful

of companies like some startups, you know,

obviously OpenAI and Anthropic, they're bleeding edge

and leading into these kind of AI adoptions.

And even then, a lot of their features are still in just like

preview mode, like cloud code design is in preview mode.

It's a really fantastic tool, but it's still not,

you know, they're still haven't figured out all the kinks yet.

So my point is that don't be afraid. Don't be afraid of all of

the large demos and like bunch of people posting stuff

into their feed that it's just demos. And like, you know,

a lot of PMs and grifters are just like

making

a bunch of posts about nothing, you know, saying that

they built something in one shot. Like those are

just demos, you know, you should really look for things

that have lasted and that has been in production for

a while. And it's like have wide usage

like Codex and ClockCode

are great examples of AI tooling that has survived.

In my opinion, all the other AI tooling. AI tooling and all

that little stuff they're just demos still they're not quite

there yet even if their valuations like crazy we don't know if

they're gonna survive so don't buy into the hype

uh forget about the demos and just look out for things

that are really making a difference in production all right

so the next tip is not judging your insides with someone

else's outsides you know this is a typical like Instagram real

highlight stuff

but basically everyone on LinkedIn or threads or x they're

all posting about kind of the shiny things and you know

they're making a ton of progress everyone

in their mom is has a new skill new clock code skill that is

changing their world and changing everything so

but the reality is they're probably spending a ton of time

iterating with clock code prompting updating the

systems and these skills probably don't work 100 of the

time they're flaky so don't really compare yourself to any

of them you know like some github that just gets tons of

stars all of a sudden. There's an interesting take where

GitHub stars don't necessarily equal to excellent

engineering anymore, in my opinion.

It's just around marketing now. GitHub has turned

into marketing

and whoever can get the most stars

the fastest because of some

hype cycle

are now the best engineers, I guess.

But I don't believe in all of that.

Just focus on yourself, focus on getting better,

and don't compare yourself to others. Self to others i

found this little cool mural over here

yeah i thought it was kind of fitting for this because

you know

she's on the outside like k-pop stars have like very

you know appealing appearances but inside she was

like struggling

but yeah don't compare your insides to someone's

outside all right so i'm here i'm eating some bagels

yum

really good bagels and the next tip tip is that you are

not behind

I think a lot of people especially when they're young in

their 20s or 30s they're always worried that they're so

behind there's so much to learn

and while that is 100% true I think when I was first starting

out becoming a software engineer I remember feeling

really overburdened like there was just so much to learn

but I myself

became a software engineer when. I was

28

I did a boot camp when I was 28 right when I got married

and then it's been like i don't know like 10 years or so since

then you know it's it's crazy i remember at the time i felt

really behind

and especially when i joined like my first company i

remember like everyone around my

level were new grads and things like that and i felt

behind but

i'm here to say

that if you're especially if you're in your 20s and 30s you're

not behind you can get started now and just learn so

much and just be able to progress now

the later tips we're going to get into kind of the nuts and

bolts on how you can leverage some of the current

systems and kind of the current

landscape so that you can stand out and things like that.

But I think it's still very possible right now if you wanted to

get started

to be able to switch careers or get into this like

tech industry, even with AI being so dominant. Now one

caveat I will say is that like if you're worried that like AI is

going to take. AI is going to take your job.

AI is going to take over everything and all the jobs.

Like if AI can truly get rid of a hundred percent of like

engineering jobs or tech workers jobs,

then what other job is actually safe? There is actually not

that many things,

in my opinion, if AI can truly take over everything

that is safe under the sun.

So why not do something that will actually help you learn

these AI tools? You know, software engineers,

in my opinion, engineers, in my opinion, are the ones that

are the best at these tools because they just inherently

understand what's going on under the hood. But yeah,

you're not too late. Now, the next tip is probably my

favorite tip, and it's that

fear is a crowded trade. This term is kind of used quite a

bit during like stock trading. You know, Buffett famously

said along these lines of,

you know, you want to be greedy when others are afraid,

and you want to be afraid when others are greedy.

And this directly relates to people essentially quitting CS

because they're afraid that AI is going to take your jobs,

people not getting into tech, people just fearing that all

white-collar jobs will be replaced and automated away

by AI.

And

honestly, in my opinion, at least with the at least with the

current architecture, I just don't see this happening.

I think AI is like essentially a super genius and also very

stupid at the same time.

It needs constant guidance. It needs good validations to

know when something's right. Like it has no taste

in the matter.

And while I do believe that a really good engineer who

knows what they're doing can 10X or 20X their outputs

by using the agentic tools. Tools but a lot of these people

who are saying that AI is going to replace everything they

just don't realize that it's not really realistic for

non-engineers or not very or people who are not very

technical to get very far

with the usage of these tools so in my opinion I still think

CS is extremely valuable to learn and it's totally worth it

and I think right now is the best time to best time to

actually learn these things and learn it well learn the

foundations learn data structures and algorithms

learn

learn system design and all of these foundational things is

going to help you become a better agentic engineer in

the future so yeah

fear is a crowded trade so avoid that don't be afraid and

optimize for learning i'm in one of these korean exercise

and exercise machines I can do like full loops

but

the first tip on this new job this new type of engineer is for

you to own the problem and not the code itself I just

talked about how the code

is losing value

but in my opinion the domain knowledge and all the

expert like problem understanding is still going to be

extremely valuable and in my opinion when code

becomes becomes cheap i think ideas and very high

quality ideas like good ideas become extremely more

valuable i think previously historically we've always been

like and resource strapped that's why like a

modern engineering

team

is like 1pm one designer and like 12 engineers for example

and like all the other support roles are really there so that

you could

select the right ideas

to build okay i'm turning here and you know famously like

steve jobs even said before that it's just as important and

maybe even more important to learn what ideas to not

pursue as well so if you want to become irreplaceable

or more

valued in this type of work is for you to deep down into a

specific domain and start owning that problem space and

trying to understand it deeply like to your root and not

just rely on like can you do this can you generate this code

or that code and things like that so own the problem not

the code so this next tip is that

taste is the mode i think anyone who has used these ai

coding tools or chat gpt or

even comfy ui for generating images

i think everyone can admit that to translate taste

is extremely. There's a reason why it's so obvious when

something is

AI generated, like whether that is writing, whether that is

an AI image, of course, things are getting better

and better.

But to me, it's still very obvious when something is fully

AI generated. Now I'm always on the sense that you

should be using these tools to elevate your work, but you

shouldn't be

completely 100% just

not thinking and letting these tools do all your work, right?

Your work right the most human part is the taste how do

you encode your taste so that the machine can

understand it and that's kind of part of agentic

engineering and we're gonna get into that but I

personally think

that

taste is a factor like let's say we're trying to generate

videos like there's zero shot that this combination

of things would happen like

for example

me coming here and

doing doing this thing

doing this thing

and then talking about AI I just don't think that is

possible and this is taste and whether or not this is

actually a good idea

or not

but personally I think this kind of things is important it

makes things more genuine it makes things more

human and

for engineers I think you need to learn how to build your

taste and whether that. And whether that taste is

about coding

or like how certain things should look or feel or design,

whether that's UX,

all of these things, you should actively study the greats,

study the work that inspires you. And then you have to

slowly start building your own tastes. And that's where the

real moat is going to be. All right, so this next tip is around

agentic engineering.

And honestly, I think it's one of the most important, a new

type of like expertise. Expertise, Andrzej Karpathy has

often talked about agentic engineering and these kind of

like learning how to use these tools effectively as a

skill gap. And, you know, Peter Steinberger talked about

agentic engineering. He has joked about the past that he

does agentic engineering during the day and around

2 a.m. He just vibe codes. So what goes into

agentic engineering?

So there's a lot, but at a high level, I think is

context engineering, agentic validations, agentic tooling,

building tools for the agent, and also like

compound engineering. And these are just some of

the pillars, I would say, of agentic engineering. I actually

teach about it in a course that I do, but at a high level,

I think agentic engineering is

a very important concept that you need to learn. I think

like as the models get better and better,

all of the sub work around the meta the meta work of

getting the agents to do more for you on your behalf is

going to be the real leverage and the real work

and all of the things i talked about of taste being the

mode and all these kind of things is going to be very

important for like validation loops right so like to teach

the agent what good is like you first need to know it you

need to own the problem right you not the code and you

need to like understand the taste and like what makes that

validation work and like how do you encode the taste and

so that the agent can recreate some of your decision

making so a lot of that is regarding agent engineering i

think it's a topic that people should not sleep on it's one

of the most important like subcategory of work like

new types of work that you're gonna everyone's gonna

need to do whether you're an engineer or pm or a

designer you're gonna have to leverage and learn these

kind of things to be able to perform at like a new. Alright,

so the next tip is around shipping things

that last. I think your reputation as someone who can ship

things that are long-lasting,

things

that last the time and has a high quality bar is gonna go an

extremely long way, especially in the day of AI. I think

because AI coding is very cheap, people tend to ship a lot

of code

and not double check a bunch of things. Check a bunch

of things. I think there's like an ongoing

kind of conversation going on right now on whether

or not

like you should review your code. I'm still on the stance

that you, if you're gonna land something to production,

you probably should review the code. This may change.

AI is getting really good at code, like code reviewing.

And sometimes it might be a code smell, but at the end of

the day,

you are responsible for your own code.

So you need to ship things that last. I think Buffett

famously said that reputation, reputation, like your rep,

takes like years and years, like 20 years to

gain and like to

get to a point where everyone trusts you 100%. But it only

takes like five minutes to

lose face and lose that reputation. I've seen a ton of

examples of this where people are just like

losing reputation for sending like AI generated emails and

like trying to land code that is like garbage because and

then just blaming AI for it. And I don't think you can blame

or should be blaming AI for it. At the end for it at the end

of the day you own your own thing

so

yeah don't be an idiot ship things that last all right so this

next tip is around following the cost of

being wrong so the concept here is that the more

expensive

it is to be wrong about your certain tasks

the less likely or the higher the bar the ai automation

needs to be the quality of the ai needs to be

for that work to be fully

replace. And that's kind of what you want to focus on.

Look for problems where the cost of being wrong

is high. Then it's less likely in the longer time that you have

before the AI fully is able to

replicate that work or do that work. So as an engineer,

you really need to look out for these kind of problem

spaces and type of work where it's not trivial for the AI

to do. You know, like AI is really good at making front-end

code these days, making landing pages and things

like that. That's like that that's not what you should be

focusing on you should be focusing on things that is hard

for the ai to do and also if the ai would get that wrong it's

expensive and that's why i like a lot of like doctors and

lawyers and like these kind of like traditional old school

things with

regulations have still haven't had deep penetration with ai

i mean there's remnants of it but it's not

as deep because the cost of getting this

decision incorrect

is extremely high so that's like a high level of thinking

when you're trying to decide on what kind of domain you

want to pick or what kind of problem space you want to

look into. You want to see

if the cost of getting that answer wrong by the AI is

extremely high. Then you should have a lot longer time

before the AI is capable of doing your work. So the next

tip that I have is a little controversial, but

I would say

you need to aim to become the top 10% of

whatever career, whatever job that you're trying to do.

You need to aim to you need to aim to become basically

the top 10%. Aim for top 1% if you can. The reason why this

is important is because CEOs and companies of all

companies

are somewhat disconnected from reality. You know,

there's some CEOs like Jensen Huang from NVIDIA

who will

never fire anyone. You know, he even famously said

that he would rather torture you to greatness than

fire you. So there's people like him who I really admire,

but there's a bunch of other companies not named

that just lay people off because of end reasons. And for

you to

kind of get through all of these turmoils and ups and

downs is to just be so good that you're undeniable.

You have to be like top 10% or top 1%. Easier said

than done, but you should be aiming for that. Like if you're

doing well, keep doing well, don't like take it easy.

As unfortunate as it is, the current environment is that you

just have to be the top 10 to be as safe as possible from

like any layoffs and even then that's not guaranteed that

you're going to survive these kind of like changes in

business needs and Etc all right so the next tip is around

learning the layers below learning the lower layers

now the thing about AI coding and these agentic tools is

that they're inherently

new abstraction layers they're really like adding

additional abstractions

and previously you would to understand the code you

have to like understand the architecture basically you

would have to understand a ton more than you are

currently allowed to understand ai is really good at

exploring the code giving you the architecture teaching

you the code being able to be productive

without you know fully understanding every single line of

code and inherently that is abstractions even engineers

before have used tons of abstractions like any framework

that you've ever used like React, Vue,

even Kotlin,

even like Jetpack Compose and like Android, those are all

just abstraction layers. Even the programming language

itself is an abstraction layer. So the issue is with AI and

these toolings getting better and better, that abstraction

layer is going up and up and up. And one thing about

abstractions is that abstractions in itself, like all

abstractions are inherently leaky.

And what that means is that sometimes the fundamental

foundation layers so things below

the abstraction layers they may break or degrade or

something can go wrong now if you don't know how to go

in and debug these layers or even just have a base

understanding of what's happening under the hood then

you're going to lose control of your projects you're going

to lose control of the work that you're doing and you're

going to get stuck at some point and this has happened

all the time for people who doesn't understand the

foundations of agentic coding or just building things

with AI,

they always get to production and then they run

into issues, whether that's security or whatever,

they always run into issues. And this will keep happening if

you don't learn about the layers below.

Now, I'm not saying here to go and, you know,

dig into every single library,

every single framework, but you should be curious as like a

default stance

and dig in where if you don't understand how things work,

understand how things work you should try to understand

it and go deeper and deeper and learn those layers you

know the deeper understanding of the foundational stuff

you have the hard things the better it's going to be for you

all right so the next big tip is around actually reading more

code now

a lot of people are going to tell you that you need to stop

reading the code because the AI is better at reviewing

the code

and the AI writes too much damn code

so it makes sense logically that you don't want to be the

bottleneck so

that just means that you have to read less code and ship

more code

but personally for me I think learning is actually the bigger

and more important thing that you need to do especially

if you're in your 20s or your 30s if you're trying to build a

good foundation you need to learn a lot of code a lot of

freaking code in my opinion

now even for me. I spend more time I think reading code

than actually writing and generating code of course code.

Of course, I'm like generating a ton of code, but I'm

hyper-focusing on what I need to read. Now,

the important thing here is that you shouldn't read

everything. There are important things that you need

to read, and there are things that you should be okay with

just letting the AI

do. Now, how do you decide what you should be reading?

Well, number one, you should read the hard stuff. What are

the patterns, the system designs that are hard?

And you wouldn't normally learn how to do that on your

own unless you are like a genius or something but a good

example is like if you want to build like an agentic tool you

should probably know about the react loop you know like

what does that look like essentially at a high level it's like a

basically a while loop

where the step one the agent reasons it hits the model it

asks if it should use any tooling based on the user's input

original input and then it acts

and then it like cycles through that over and over until the

agent finishes so like that code that piece of code that

makes your application agentic is something that

you probably want to learn about so maybe go read

open code

you know deep dive into that and then see how that open

source code is actually made and how tools should be

organized so all of these things the hard parts the things

that matter is what you should be reading and all the little

stuff like how to center a div or go i

don't know make some random react component.

These are like easily verifiable these days with like unit

tests or component screenshot tests and I don't think it's

worth that much to read unless you have no idea how

those are done.

So as you learn more and more about coding and how

certain things are built under the hood the less you need

to read that specific code right and leverage kind of the

meta agentic gardens like tests and validation loops

and etc

to get more out of your agentic coding. To coding so in

my opinion you still need to read a ton of code and I think

it's more important to just like read a lot

and learn a lot about these foundational things anything

that you don't just like understand you should read about

it and learn about it and then and try to build it and

use the.

AI to vet your understanding so read a lot of code all right

so the next important tip is

on ignoring titles and focusing on shipping when I first

became an engineer I actually just tried to ship a lot of

things and i remember that really helped me move up very

quickly and i think that is still the same these days where

you just have to focus on building and shipping like

real valuable

products you know and just don't let your title like hold

you down i think that's more important than ever i think

the best ideas will always win so if you're in this position

where you are like new to a company don't let your

position hold you back you know use ai use all the tools

that you have to like really get a leverage and just build

your ideas and test it out all right this next one is one of

my favorites and it's actually learning where the ai fails

so the thing is if you use these toolings enough you'll kind

of eventually fall into this zone where you inherently get

the second sense of knowing what the ai can do or cannot

do so you start developing the sense of like what these AI

systems are really good at and bad at.

And I think it's actually a skill in itself to have this like

second sense. It's like problem solving in a sense.

If someone says, hey, I need to do this kind of automation,

what can you do? How can we do this? Then you should

inherently just know immediately

like, oh, we could probably use this kind of data extraction.

We could probably have this kind of context gathering.

And then we could have like these kinds of AI systems.

AI systems in place that will do the work and automate

that work. So that whole workflow is learning what the AI

can do and what the AI can fail. And one of the most funny

things is, if you think about it, all LLMs can do is really just

hallucinate at the end of the day. Everything is just like a

guess that the AI

model is making. The ones that are useful just happen to

be useful be useful hallucinations, right? So get a really

good intuition of learning when AI can fail. Like start a log

or like just

start keeping track and try really pushing as far as you can

with the systems and test where the AI can do something

and cannot do something. All right, so the next tip is that

you need to have a T-shaped portfolio. Now, I made an

entire video about this earlier in earlier in the year, but the

high level idea is that you need to have essentially a wide

depth of knowledge on various things because this will

help you know what is possible

with the AI, like what you can physically do

and quickly navigate from different projects to projects

and learn different domains really quickly. That's kind of

the high level idea that you want the breadth of

knowledge so that you

can coordinate and orchestrate multiple agentic tools

at a higher level of abstraction.

But you also need at least one really deep understanding

of something foundational. Like you're not a specialist

per se, but you just need like a deep

understanding of at least one tech stack. And that will be

like your main work that you do at like work or whatever to

be competitive.

But then you just need to have this like breadth of

knowledge on top of it so that you should be able to pull

from these different sources and different disciplines to

perform even better at your current job. So you need to

have a T-shaped portfolio and this kind of portfolio is

what recruiters are currently looking for. All right. So the

next one is on mastering AI tools.

And I think this may be one of the most important tips I

would say. Like you just have to learn cloud code or

codecs like the back of your hand. You need to

understand how it loads memory,

how it manages context, you need to understand like kind

of the important slash commands and skills, what a skill is,

what an agent

is, how to do multiple sub agents, agent teams, how to like

teleport your instances, how to do remote control, how to

do scheduling loops.

I know I just listed a bunch of stuff and it's maybe

like too much.

And you may wonder like, do I really need to know how to

do all this stuff? And in my opinion, I think

yes, you need to know these things just like a professional

woodworker knows how to use all of their tools efficiently

and professionally to be able to make like complex

joineries and et cetera. These agent tech tools are the

tools of the future and you need to know it. Now I would

also go as far as. Other AI tooling, like even

video generation, image generation, like which AI,

which image model is the best and for

what specific purpose, which video model is the best and

for what purpose voice models and how to do like local

LLMs using a llama.

And there's just so much that you can learn in this new

area of tooling and agentic tooling. And in my opinion,

if you really want to stand out and be great in this new

environment is to master these tools. Just be able to

leverage these tools alone.

I think can get you hired. All right. The next tip is the world

best tutor is an AI. I personally find that I can learn

basically anything that I want with AI, whether it's

to learn blender, whether it's to learn Adobe,

a new software,

or just doing a deep dive in some code base, like looking

into open code or learning about TPUs versus GPUs.

Whatever it is, I'm using AI

constantly to learn. And I think the important thing here is

that you develop a desire to want to learn. All right.

The next one is pushing it to prod a little shout out to

my newsletter, get pushed to

prod on sub stack. But this one is really about pushing

through to shipping something all the way to the end.

The number one thing that happens with people who start

building stuff with AI tools is

that you just chase the red dress, meaning there is a

bunch of new shiny things, new side projects all the time.

So. You do this, a little bit of this, a little bit of that,

and then you end up never shipping anything

to production. And in my opinion, you learn the most if

you ship something all the way from

like zero

to end and shipping it and then maintaining it.

And you just don't learn the same lessons. And in

my opinion, the people who are very senior, the people

who are way at the top

have shipped a ton of things and have the scars to

prove it.

And because they have the scars to prove it, they're able

to handle situations.

That most

engineers who have never shipped anything to

production and maintained it for more than a year

or three, they just don't have the depth and knowledge to

handle certain situations that will

inevitably come up.

So push yourself to ship it to prod.

All right. So this next tip

is around communication is the largest leverage that you

will have.

In my opinion, being able to properly articulate your ideas

is still the killer skill that most people lack.

Now. Unless you're like Carmack, who's like a genius

engineer who can essentially build the next graphics layer

that the entire industry

adopts, then you're kind of shit out of luck.

You got to do what everyone else does.

And in my opinion, being an amazing communicator will

greatly help you stand out amongst the crowd.

And I always say that like doing YouTube for me personally

has always started from a place of trying to get better

at speaking. And if you ever want to see

like someone suck at communicating, look at my first

ever video

and kind of the evolution of my journey on YouTube and

because I do YouTube prolifically

and I treat it as a skill like anything else, and I try to

improve on my speaking and the way that I deliver

messages and things like that. Whenever I have to give a

presentation or do anything of that matter, I'm never

afraid to do it because I just know that I can perform

and execute on

this. It's kind of a verbal communication skill. And in the

age of AI, when there's just so much noise, I think to be

able to stand out, these are going to be the soft skills that

there really isn't a place for AI to really interject here,

maybe help you script certain things. But you know,

like I said, you don't want to sound like an AI.

You don't want to sound like

you're reading off an AI script,

right? That's that will ruin your reputation, right?

So I, I still believe communication is going to be the single

most leveraged skill that you can learn. Besides everything

else that I mentioned,

right?

Now, this next tip is around becoming worth vouching for.

Now, in a lot of my videos, I've always mentioned that I

have optimized for people in a lot of my careers.

So what that means realistically is that I found strong

leadership and I kind of stuck with them. But at the

same time, I worked incredibly hard to earn the trust of

my leadership, and I essentially became a person that

people would be very happy to vouch for. And I've done

this with every manager that I've ever had, starting from

my first manager in software engineering. I still chat with

him from time to time. But

the thing is, the industry is a lot smaller than you think.

For example, in Gemini right now, one of the main VPs

from Instagram went to Gemini and a lot of people are

going over there to Gemini as an example. And I'm sure

like OpenAI and Dropback has a bunch of meta people

and people are going there.

So the thing is, you need to have good reputation and

these reputations will last a long time and to get good rep

and to be vouched for is you to build it it's essentially just

all the things that i talked about so far and the rest of

the video

essentially but yeah so you always want to put your best

foot forward and eventually and over time you'll just build

this reputation of yourself and people will know about you

people will want to refer you and you know referrals are

always the best way to get hired so the next tip is to do

feels like play

and in my opinion with all of these tools sometimes it just

feels so magical

and it's actually quite fun

and I talk a lot about this in a lot of my clock code tutorials

but sometimes when I'm doing a multiple agent

orchestrations with multiple panes and I'm just juggling a

bunch of clock code instances it kind of feels like

Starcraft like I'm playing. Starcraft and there is a sense of

like joy that I get and I know this kind of experience is

probably not enjoyable for everyone you know to each

their own i say right but try to find what is fun and you end

up finding a lot of enjoyment and satisfaction from doing

it there's this famous quote from steve jobs it kind of goes

like this where he was saying that the people who succeed

in life in long term are the people who ended up finding

something that they really were passionate about and

then stuck with it for a really long time because if you are

not that passionate about it or you don't find enjoyment

in the thing that you're doing then when it gets hard or it

gets frustrating you're gonna end up quitting and then

the ones that are

really crazy about the things that they're doing they're the

ones that stick it through during the hard times and the

long times so find something that is essentially play for

you so this next one is following the money and it's not

actually what you think it is and

the thing is if you're just starting out chances are getting

into one of these like major ai labs or fang it might be a

little bit out of reach unless you have like really good

internships and things like that it just is kind of hard to

reach so in this case you want to just kind of follow the

money the funny thing is a lot of these ai tools that are

heavily being leveraged are being done in a lot of like

traditionally boring industries you know like finance or

farming or some bespoke place might just be the ones

that are leveraging AI the most.

So look out for it, look out for those things and then try to

get into it. And that way you can essentially be in a place

where the demand for AI is just super high and there's just

not a lot of people that are looking into it. So you're

essentially looking for a place that is not crowded and you

could find that by just following the money. Now, the next

tip is taking risks before you can't. And I actually did this

myself when I was 28. I quit my job and went to a

boot camp. At the time, it was kind of a big risk. I didn't

have that much money. I wasn't making that much.

But yeah, I went and did the boot camp and it paid off

quite well. And the thing is, this is very true right now.

I feel like doing a startup, optimizing for opportunities

for learning.

I think these are the risks that you should take, especially

if you're young, because at some point it'll be harder and

harder to take risk. And,

you know, for me right now, because I have two kids and I

have a good paying job, the risk of doing something else

or quitting my job or, you know, pursuing. YouTube or

doing just something else than what I have going on is

huge risk to me financially and also like security for

my kids. So if you're young, take risks. And many,

many famous people have said that when you get older,

you don't really

regret the various failures, but you definitely do

regret the risk that you didn't take for certain, certain like

new experiences or certain opportunities. So take the risk

that you can right now. All right, this next one, you want to

optimize for slope and not your salary. Now, when you're

first starting out and you're making a change or taking

a risk, you want to change the slope of your learning.

You want to maximize whatever opportunities that you

can have to increase the learning as much as possible.

The salary and all that kind of stuff,

eventually

comes down the line and you'll be surprised that if the

slope that you created is

steep enough, eventually in a few years, the amount that

you'll be making difference will be made up completely.

I remember when I was first starting out, I was kind of like

comparing between job A and job B worrying about like

20 grand or

whatever. And I was like really torn because I wanted to go

to company A, but it was paying like 20 grand, a little

bit less, but it had like a better learning opportunity.

And I'm so glad I took the learning opportunity because I

think that elevated me to go

further faster.

And now at my level, $20,000 doesn't seem like anything

at all. In fact, my bonuses end of the year is multiple times

that amount. So you don't want to optimize for like little

bit more here or there, but you want to really look out for

the things that will elevate your learning as fast

as possible. All right. So this next one is about owning

the ugly work. Now there is one caveat to this is that even

though the work may be ugly, not desirable, it has to

be important. That is one caveat for this tip.

But the thing is, especially when you're

early and you're starting out, you may often see that like

the best quote unquote best work and the most shiny

work is given to the more senior and the people who are

more established. And the thing is a lot of junior folks tries

to figure out a way to work on those projects,

which is good. But at the same time, it's crowded.

When the work itself is crowded, the large portion of

kind of the merit goes to the leads and like the

senior folks. So instead of going to a crowded space,

if you can, it would be better to own

an ugly piece of work. Maybe it's insights, some migration

or some core piece of technology in your organization

that is still valuable and very useful and critical. But it

happens all the time that there's just things that people

end up not owning, even though it is critical to the org.

So doing the dirty work, doing kind of thankless work and

then doing it consistently and doing a good job will

definitely help you move up quickly

if you do it right.

Now, there is a caveat that you have to be vocal here and

you have to make sure that it is actually still valuable.

Don't get confused of like

ugly work that is like also not valuable. Then that's like

something that you should definitely avoid. Now, the next

tip is for you to become the bridge between your

company and

the AI products. So

these tools, in my opinion, there's a clear skill gap.

And right now, most of the people that I've talked to,

there are a lot of engineers that I've talked to that are in

different levels, like principal engineers, architects and

senior engineers. And you know, I've talked with a gamut

of engineers. And the thing is,

almost universally,

most people are not sure what is quite possible and not

possible with these AI tools, unless you have spent a ton

of time investing in cloud code or codecs. The thing is,

I have. And because I've done that, I've at least in

my company, I was lucky enough to be in a position to be

able to lead a lot of our like AI transformation work.

And that's kind of the key. A lot of companies right now

are doing.

AI native transformations. So what that means is if you are

the person who knows the most about these tools and

what is possible, what kind of integrations is actually

doable and at what scale and how much it's going to cost

and those kind of if you can answer those questions,

then you're going to be in a really good position. Now,

if you're just starting out, you might not be the lead

of everything, but because

all of this stuff is so new, even if you're like a new engineer

or like a senior engineer, you might be able to be in a

position that is leading these efforts that usually maybe a

principal engineer would usually lead. But right now,

because all of this is so new, those engineers who's

been around, they're just not going to automatically know.

They're going to actually have to spend the time and really

dig in and get the ground source of truth by just being in

codex or clock code for like 10 hours a day for like a

decent amount of time. And most like principal engineers,

they just don't have the

time because, you know, they're in meetings most of the

time and a lot of people aren't weren't coding really. So in

my opinion, some of like engineers who are just like

investing all their time learning these things are going to

be in a much better position. So be the bridge between

your business and the AI. All right. The next tip is around

networking before you need it. In my opinion,

networking is an extremely important skill that you

should learn, especially if you're early in your career.

I often say that opportunities are attached to people,

not specifically job boards. And you can see this in action

if you ever went through like an referral experience

versus just

cold applying. Now, in my opinion, the best way to actually

network long term is to just have genuine experiences and

work colleagues over time. And that's the best way to

build your network. So getting into a really good tech

company really helps a long way because in these

larger companies, people cycle in and cycle out. And

in the in like a few years of working at like Meta or Google

or any of these like major tech companies, you'll

one day wake up and your network will be amazing. It will

just be all over the place. But if you don't have that,

you need to

somehow create that. And in my opinion, there's a lot of

ways to kind of build that network without any of that

as well.

One interesting way to build your network is actually

through social media and

building content. Now, I make content around tech.

Since my time making content, I've found a bunch of

content creators that talk about tech and they are now

part of my network. And I try to be as genuine

as possible. I'm not just making these contact points just

so I could ask a favor or whatever, but it's really just

because I'm genuinely interested in these people's work.

And if something comes up where there's like an

opportunity to work together, that's like just like a cherry

on top. So I highly recommend you to build a network

before you actually need it,

because when you don't actually need it is when it's the

easiest and the most genuine for you to build a network.

The next tip is on winning the interview. Now, I think

interviewing is a skill. It's sucks, but it's a skill that you need

to learn.

Now, one of the things that people often ask me these

days is that because of AI, has the interview changed?

And I think the answer is essentially yes and no.

Fundamentally, the interview landscape is changing and

has definitely changed quite a bit, but not so much,

honestly.

For example, I think lead coding is still valuable. A lot of

companies are still a little behind when it comes to this,

and I don't think they'll change the way they interview,

at least for another year or so. Who knows, really.

But fundamentally, lead code or these kind of like lead

code style questions where maybe they set up an

environment and you have to like solve or fix

some problems. Those manual coding assignments,

I think, are still going to be around because honestly,

there hasn't been anything that's like more

like kind of a direct way to know whether you know how to

code or not. And then system design is still going to be a

large portion. If you're starting out, it's probably not

as big, but I think it's definitely worth

learning system design deeply these days. I think because

of AI, I think that level of abstraction is going to be even

more and more important. I think it's going to be graded

even more heavier. Now, there is a new type of interview

that I think it will like evolve and be like created.

And one of them, I think, might be around AI coding

on essentially agency engineering, but not enough

companies know about it yet. Like they just don't.

Fully understand the concepts, but I guarantee in like a

year or two, knowing how to use these tooling because it's

such a critical part of your job is going to be part of

the interview.

I could almost guarantee that. And I think there's some

remnants of it already where there's some

places that have AI assisted interviews, but everyone's

kind of figuring this out in my way. The best way to still win

the interview is honestly, it's still kind of the old traditional

legal data structure in the algorithm. System design,

behavior, interview practice,

doing all of those, nailing all of the fundamentals.

And then for each company that you are applying,

making sure you brush up and try to gain as much

information on any new processes that they have.

But again, I still think getting into the best tech company

that you can is still the best advice I can give you.

Either that or join one of the very hot startups

that are very well funded and is in the bleeding edge of

AI usage. All right. The next tip. Next tip is around picking

the manager and not the logo. I think your manager is

probably the most important

person

that kind of determines how enjoyable your job is and how

well you can actually do.

A good manager is really kind of a game changer in so

many different aspects.

They fight for you in rooms that you're not part of, which is

like a very important and under rated thing.

The performance review cycles and things like that,

if your manager is your biggest advocate

and without his or her help, you will not be able to

succeed in your organization. A lot of people like

understate how important a manager is. And the only

thing is just a performance thing. But there's so many

things that a manager does behind the scenes that you

may not be

aware of. You know, they may protect you. They may

protect your time.

They may guide you and mentor you to pick the best

projects and things like that. But one caveat here is that

you need to pick the right horse. You need to know how to

find good management. And

one of the meta tips that I have around this is that you

need to look at your manager and then their manager and

then their manager and then see

how strong that connection is. One easy way is to just

ask or figure out how long they've worked together.

And then another really good sign is how many very

strong ICs want to work with this manager. Because all of

these signs is that this manager is successful and is a

good leader. And in my opinion, having a good manager

that you could work with for a long time and getting onto

the bench is something what I call that will pay

you dividends

long and long. And not only in terms of your

career growth, but just like your sanity and having a good

manager is

worth its weight in gold, in my opinion. All right.

The next tip

is on finding great

mentors. Now, through my careers, I have had many great

mentors and I feel like all of them has saved me

many years in my journey to get to where I am today. And I

feel like I have mentors for a lot of different things,

whether that is like deep dives on performance,

like technical mentorship. And there's also mentorships

on how to navigate your career.

And

also like even YouTube, I have like mentors

around YouTube. I feel like having great mentors will save

you a ton of time. And I think in the age of AI, where

you're learning the taste and learning what good is, I think

having a mentor will really help you level up quickly.

The thing is, there is a limit to what AI can teach you. It can

teach you like general like knowledge and like deep dives

and things like that. But

when there's too many

choices, the AI may glaze you. So you really need a mentor

to kind of like guide you and course correct you

throughout your journey to becoming an engineer.

All right. So the next tip

is to teach to learn. So

I often say that you don't really know if you actually know

something or some topic until you

try to teach it to someone else. You really only gain true

mastery if you could teach it to like a beginner. Now,

of course, not all great engineers are great teachers,

you know, not all players in like sports are great teachers.

I myself have trained hundreds of engineers,

And this is not like an arbitrary figure. I've actually trained

hundreds of engineers

on cloud code and codecs.

I actually teach a course on the side for fun. But also

internally at my company, I've led multiple sessions with

like hundreds of people in it. And I taught them

agentic engineering, cloud code, and just like all of that.

And I've led multiple initiatives on this topic.

And I think the only reason I was allowed to even be in this

position to

like teach is because I learned it really deeply. And I like

advocated for it very heavily. And I initially just started

making videos even internally about the tooling.

Just teaching everyone everything I knew about

these tools. Because of those teaching, I found

just more opportunities to teach more people.

And because I am teaching more people now, I am

spending more time

myself deep diving. Whenever a new feature comes out,

I want to learn it because I want to teach it. So there's this

kind of like

cycle and flywheel that happens when you are really

into teaching. So, yeah, give teaching a try. I think you'll be

surprised how much you don't actually know something

until you try to teach it. And then your students and

people who you teach from will ask you questions that will

challenge your understanding.

So teach to learn. All right. So the next tip is on picking a

direction and sticking with it. When you're first

starting out, there's so many distractions. There's so many

like red dresses I like to call.

And, you know, you want to go work on this. You want to

work on this side project. You want to dive into this thing.

And if you start too many things and never finish it,

never ship anything to production, then you kind of lose

out on progress.

You you'll end up kind of like mistake motion for progress.

And that's like not the best move when you're trying to

get something done.

So in my opinion, you need to pick one thing,

whether that is like learning mobile development or

building some agentic tool, like see it through. See it from

zero to end all the way through. Pick one direction and go

as deep as you can. And going back to kind of the

T-shaped portfolio, this will help you get that depth by

building deeply and picking one thing. One of the

common things that you see here is let's say that you're

climbing a

mountain and if you climb one mountain,

you'll like go to different peaks and valleys right of

that mountain.

And imagine those are like roadblocks or like challenges.

And at some point, you'll hit one that is hard to do, hard to

get past. And because of that, you'll just go to a different

mountain and start from the beginning and you'll feel like

you're getting a lot of progress because the initial

problems are probably very similar and you've already

experienced them before.

So you're not really learning. You're not really gaining

any momentum.

Instead you're going to hit that same like peak,

something similar. And then you're going to

want to go to another mountain because the

new learnings and new challenges are hard and the gains

are like

further and further in between. So avoid those things.

If you really want to see progress, you need to pick

one direction. And then go really far. All right. So this next

one is

about building in public or learning in public.

And personally, I think it takes a lot of courage to do

anything in public, like even filming this and because

you're going to share

things that are maybe not quite ready and you're,

maybe you're a little embarrassed to buy it. It helps you

build like tough skin. It also helps you want to

like finish something fully so that other people can check

it out. There's just so many elements to building in public

that I think is good for

people that are trying, trying to start out.

Now the other besides like the self like improvement

aspect of it, I think there's also this angle of being able to

network very easily, right? You're going to be able to find a

lot of people that are interested in your

things. And you know, sometimes you might even just get

straight up, get a job off of it or

like in cases like open claw,

maybe you make like an open source project that goes

crazy and everyone knows about it. And now you're one of

the staple developers in this new AI ecosystem. So you

never know what's going to happen.

So in my opinion, like building and sharing in public is a

extremely valuable thing, especially early in your career.

So yeah, building public and learn in public now

continuing on this topic. I also think you should really take

a moment to audit all of your expenses. You'd be surprised

how many subscriptions you probably have that you

don't need. You really should just focus on the core things

that you need

to essentially learn what you need to learn, survive,

you know, eat and things like that and save the rest.

But

yeah, just, um,

audit every single thing that you spend on

like your audit and code for example, you'll be genuinely

surprised how many things that are probably slipping

through the cracks. So yeah, so just get your house in

order and audit your expenses. Now on the topic of your

tiny cage,

I would say the next tip is that your first 100K to save is

going to be

the hardest. There's this kind of saying that your first 100K

is extremely difficult.

And then your next 1 million is very difficult as well.

And then the next 5 million feels impossible.

But once you hit these milestones at some point,

it becomes a lot easier to make money. To save $100,000

is extremely important because it gives you a lot of safety

and wiggle room in the future. And also there has been

lots of studies that said that if you have saved $100,000,

you're much likely to

save a million dollars in the future. All right. So this next

tip is set it and forget it. I think I would feel bad if I didn't

leave some investment advice for,

like if you're especially if you're in your twenties

and thirties,

but essentially pick an index, put some money in there

every month and just set it and forget it. Like just do

this consistently. Hopefully you have built a small cage

and you don't have that much fixing expenses. Just makes

this part of your fixed expenses. This is not advice

specifically related to AI,

but in my opinion, in this day and age, we don't know

what's going to happen in the future.

So saving for a rainy day is like extremely important.

So just, just take advantage of the compounding.

The compounding effects of setting it and forgetting it

and just have a recurring, like pick some good index like

the SMP 500 and then just

invest.

This is not investment advice though.

So I've told you to save a bunch of money and build

yourself a tiny cage.

And the next tip is around what you should actually spend

your money on. And in my opinion, the number one thing

you shouldn't skimp on right now is your AI tools.

You should pay for your own AI tools if you don't have

access to it. The thing is these tools are not expensive.

The max plan for a clock code,

or codex is like $200. And depending on how fast you

use it, you could hit the rate limits pretty quickly.

But personally I pay for these. I also get a bunch of tokens

at work. But the reason why I kind of encourage people to

spend the money on this stuff is because it's learning.

As I mentioned in this video, I think these tools are the

single most important tools that you need to learn in the

next like few years. You just have to be really good.

You have to get so good at it that

you know how to debug things. You know how, you know

the layers below, the abstraction layers below on how

it works.

You know all of the skills, all of the plugins,

you know how context is managed.

To do that, you need to spend your money

where your mouth is essentially and really

dig into these tools. So don't be stingy here.

Don't wait for it. So use these tools to the best of your

abilities and use it now before they change like kind of the

plan is, you know, this $200 plan with the current usage is

not going to

last forever. In my opinion. All right. The next tip is that

rest is non-negotiable. It's kind of funny with AI.

You would assume that a lot of these things are being

done autonomously and like a lot of the work you probably

don't need to do anymore. But I personally find myself

working even more and I think part of it is because there's

a lot of people who

are like me who are like builders at heart who wants to

constantly build things. I actually have a video on being

addicted to cloud code. So yeah, I think you know,

you really need to get your rest. You don't want to just like

keep pushing yourself. I think I think there are times when

you need to push times when you should rest.

But if you don't get a habit of like working out and relaxing

and resting then you're not going to make it very long.

It's a marathon and not a sprint at the end. The next tip is

on building your escape hatch or multiple

income streams.

Now I myself have multiple income streams now like work

is just one part of it.

But YouTube is an income stream. My newsletters are

income stream. The course that I do is an income stream

and I'm always looking for new

income streams. Now this whole like section around

saving watching your costs not spending that much

money building kind of your wealth. The large reason why

I'm talking especially about this, especially if you're in your

early 20s and 30s is because of the uncertainty of

the future. A lot of people talk about the K shaped

economy and how there might be a permanent middle

class and the people who have all the tokens are the ones

that are going to. Essentially have all the money and all the

power and honestly, it's a little scary and it's a

little unsettling,

right?

So to me, I think the

best thing you can do is make sure you have a lot of

escape hatches. Make sure you have a lot of sources of

income just in case something happens to one of them.

And yeah, so I am not going to stay here and say hey,

I know the future. I know what's going to happen. So the

best thing you can do is make sure you have an

escape patch. Save your money and be able to invest it

properly so that you don't it just doesn't burn to inflation

and things like that to prepare for

something just in case AI does in fact take

over everything. Now the next tip is around

your physical health and it's really like training like your life

depends on it.

I think you need to work out you need to be fit this

desk job.

I personally found it to be extremely

like painful as I'm getting close to my 40s. My wrist hurts

my lower back hurts all the time.

I'm tight and I just wish I like stretch more and like use the

standing desk more when I was younger

and did like yoga and things like that. So these days I

have like working out and health as like a high focus.

I keep saying this but it's a marathon all of these things is

a marathon. It's not a sprint

and if your body is not healthy and it's not

in the right state that you're not going to make it very far,

You know, you don't want your body to be a limiting factor

on your growth and your success. So the next tip is to be

an early adopter. Now, personally, for me, I've always been

an early adopter.

I highly recommend being an early adopter, especially for

AI coding tools and also not just coding tools, but,

you know, voice models, image models, like the best

AI search, like perplexity, like all these like different things

that you can do with AI. I think it's worth being an early

adopter because it will let you essentially see trends and

be able to live in the future a little bit.

Now, the crazy thing about all of this stuff is that like Opus

4.5 is a really good example. When that model came out in

November of last year,

I remember

I just tried it right away because there was also like a

double rate experience during the holidays. I just

remember it was like, wow, like something clicked,

something changed. And because

I adopted to it very early and I switched all of my

workflows to it, as soon as I found that it was like

game changing, I was able to learn a lot more than other

people around me faster. And because I was in

that position, by the time that the entire industry kind of

woke up

and

around February, I think that's when like it spiked end of

January and around February,

at least for our company, the usage of cloud code

just exploded. And by then I had already been using 4.5

for months and I've been heavily

just optimizing skills, learning all the fundamentals.

So when everyone was struggling and learning, I was

ready to teach. And what I did was I took all of that

experience and started teaching people. And that helped

me get into a position

where I was able to lead initiatives and lead AI

and basically create new scope for myself. And I think you

could do this

by being early. And in my opinion, especially for

these tools, doing some investigation early and chasing

some of these tools, I think is very important and it will be

very beneficial because it will help you get ahead of the

game and ahead of the curve when it comes to

these things.

Now, a tip on how to be early and how to get informed on

all of these things is to follow the leaders.

I personally follow a bunch of people like Boris Chenry,

Andrej Karpathy, Peter Steinberger, obviously Dario

and Sam,

also Elon. I don't care about politics or anything. I try to

keep my ear on the ground and get informed on these

things that are kind of coming down the line.

And whenever someone says anything like ridiculous,

you know, when Jensen says something like,

oh, a $500,000

engineer who gets paid that much, a salary, if they

don't spend

$250,000 worth of tokens, then I'll be deeply concerned.

So when he says things like that,

I really try to think about it.

A lot of people clown on it, a lot of people dismiss it,

but I think there's some validity to that.

It really depends on your perspective of what tokens are,

like if you truly believe tokens are a unit of work.

But anyways, I have another video around tokens and

loops and things like that, so that's coming out. I have my

own AI news

source

that I built with ClockCode.

So I have people that I think that are worth listening to,

and I just have these things and I consume them.

And because I'm also making YouTube videos about AI,

I'm always trying to be up-to-date and trying the latest

and greatest

things. It's a lot of work, but at the end of the day, I think

right now is a very special time where these things, the

cycles of innovations and the competition is so fierce that

I think if you are not actively pushing yourself to be on top

of these things, you're going to kind of fall behind.

And you know, there's nothing too wrong with it,

but I think if you want to be early, if you want to

adopt early,

and you want to kind of look to the future a little bit,

you kind of have to be in the know, and especially what

kind of things the leaders in the spaces

are like really thinking about what's in their head space,

you know, because I'll give you some signs and clues to

where things may be going in the future.

So if you are able to do all of these things

and you're essentially financially free, you can make a lot

of decisions,

I would highly recommend you to bet on yourself,

whether that is, you know, quitting a job and like

switching careers, for example, that's like what I did.

I bet on myself that I could learn like software engineering

and somehow

ended up from like a PM

to become a staff software engineer at Meta. It was a

long journey, but I essentially bet on myself every step of

the way. So whether that is going for that big job or

starting your own thing,

I think at the end of the day,

this is a great time to really

bet on yourself to essentially leverage these tools,

especially when a lot of people are kind of sleeping on it,

they're not really learning how to use these

things efficiently.

If you can get ahead

and really become an expert in these things, you will get

really far, whether that is in your tech career

or like doing your own thing. All right, so the next tip is

around shipping it while you're slightly embarrassed.

So the thing about shipping products,

especially when you're working on a side project

or something, you want to build it to a point where like the

MVP is obviously built.

So the term MVP, I don't really like, like I think a product

being minimal viable is like not really ready in my opinion.

Like you need to go a little bit above that, something that

you feel pretty good about. And it's like covers a lot of the

edge cases and things like that.

But shipping something

when you're slightly embarrassed with it, especially the V1,

I think it's extremely important. Like if you're not

embarrassed by that first version that you're shipping,

in my opinion,

that means that you probably waited a little too long

before you shipped it. There's a balance of speed

and execution that you need to really think about. And if

you don't get in the habit of like putting yourself out there

and shipping things where you're slightly embarrassed

about it, then

you'll never ship anything. I met a lot of perfectionists in

my life that

like always talk about shipping things, but never end up

shipping anything because it just comes with

the territory. If you worry about all of the little details like

too much, especially for V1, then

you're never really going to ship. And also once you

ship something, you're going to get a ton of feedback and

new ideas and things

and ways to improve your product that you never would

have like understood.

So ship things

that you're a little bit embarrassed about. Actually, V1.

All right. So the next big tip that I have is that

luck, in my opinion, is when

preparation meets opportunity. This is one of my

favorite sayings. I don't know where I picked it up. It was

probably from a movie called Serendipity, maybe.

I'm not sure. But the high level idea is that you go through

life where there's just a bunch of

opportunities everywhere. And if you're not ready, if you

haven't been like training, preparing,

learning,

building, if you're not sufficiently prepared enough,

then you may not be able to seize an opportunity. You may

have experiences from time to time where maybe you

went to like some meetup or some event, or maybe you

got an interview and you just couldn't nail the interview

or you

had nothing to offer to someone when you had some luck

to meet some people. And that's why I feel like in this day

where AI is like so booming, there's going to be a ton

of opportunities.

I think it's just like everywhere, there's going to be

opportunities everywhere. It's easy to raise funding. I think

it's easy to land interviews with a bunch of new startups

and stuff like that. But if you're not ready and prepare

for it, then you won't be able to take advantage of it.

All right. The next tip is that your first job is not your

final job.

A job is just a job at the end of the day. Like if you lose it or

if you take a first job that is not like perfectly ideal, it's not

the end of the world. You'll have many jobs in your career

and

you should just take the best job that you can when you

start out. And then just keep going, you know, and then do

the best that you can. And if a new opportunity comes up,

you just seize it. A lot of people, especially when

they're young, feel like that first job is super important.

And I'm here to tell you that it's actually is like not

that important. Sometimes you could start in like a totally

different industry and then switch to another industry.

But the important thing is that you keep moving forward

and making sure that you're not just complacent in one

job and one role. You need to be constantly evolving

and growing. So just remember that your job is not like

your final job. Okay. This next tip is around if you're

building agentic applications.

I think one of the best advices that I heard is essentially

to build

for a model that is available in the future. So right now,

I think every generation of a new model release,

something that was previously not possible

becomes possible. When Opus 4.5 came out, I think that

was like a moment where coding with AI wasn't really that

great,

but that model just made it amazing. So like there was a

spike and it was like kind of a combination of Cloud

Code's maturity and also Opus 4.5 coming out. I think that

kind of made a big difference. I was at a talk with Boris and

this advice of building for the models of the future is one

of the best advice that I heard from him. And that's

essentially what Cloud Code was. He mentioned that

when he was working on it, he made it so in an assumption

that future model cycles are going to

build features with the assumption that eventually the

models will catch up. And I think this is unique for people

who are interested in building agentic applications,

because if you are just building things for the current level

of the models, then it's a lot easier to copy those models,

right? And this kind of limits your thinking to what you

think is currently possible. But yeah, so if you're building

agentic applications, I think building for the future models

is a great advice to take. All right. This next tip is around

not letting AI atrophy your

skills. I think one of the common things that will happen

when you get really deep into all of this AI coding and

AI tooling

and leveraging all of these tools is that you just

stop coding

manually. Now, I don't really think manually coding is like

that important anymore, but you should be constantly

studying like system design. You should be constantly

studying the core foundational things, like not forgetting

about space time complexity, not stopping to read code,

because all

it's going to atrophy and it will happen a lot faster than

you think. And as unfortunate as I think it is, at least

right now,

the industry is still in a position where you will need to

do decoding

and kind of like manually coding probably for

some interviews. This may change in like a year.

This advice might be totally outdated by them. But yeah,

like

watch out for your skills atrophying and making sure that

you're like putting in cycles of extra study and

continual education.

Like I myself personally have dedicated time where I'm

deep diving into various system designs like every day

having deep understanding of how a lot of systems work

and a large breadth of it is going to be really important

and

also it's really interesting right now because a lot of new

patterns are being

created because building agent tech apps is like a whole

new thing and.

I don't know there's just like a lot of interesting things that

are going to come out like agent tech voice design and

other like interesting experiences that needs to be built

on top of this like model routing and I don't know I

personally just find they're very interesting so don't let

your skills atrophy and continue to pursue greatness.

All right I hope you guys enjoyed that video I know it was a

long video 50 tips was no joke to film it took me quite a

while actually I had hoped to finish it all earlier I think at

the end of the day do not despair

I know it may look like doom and gloom sometimes and if

especially when you're young and starting out it feels like

nothing is working out and it might feel impossible

but i will tell you that there is light at the end of the tunnel

for those who work hard and push through this and and

then i hope you guys make it i hope you guys find success

and truly

am able to thrive in this new environment with that said

i'm going to enjoy the rest of my time in korea in my

vacation sorry for the delays and videos these days but

yeah I'm on vacation so

I have a bunch of AI coding videos and other things on

this channel so feel free to check it out and until I see you

guys on the next one

bye