If I Wanted a Tech Job in 2025, I’d Do This

how does one go from studying something

non-technical to being an ml engineer

for inference at Nvidia in under 2 years

hi welcome back to episode 24 of backs

of all trades after quite the Hiatus and

what a way to come back I'm here with my

good friend isan thank you for coming on

the show man thank you thank you thank

you for having me so isan is one of my

close friends um and we'll get into the

story a little bit but he has one of the

most interesting stories I've heard from

anybody really when it comes to being a

new grad and the sort of both turmoil

but just crazy Journey you've had uh and

from startups to crypto to ml to getting

acquired twice and uh I think that a lot

of people will find Value in your story

but also just your insights around uh

just doing things which I think is

something that we both share so uh I

want you to elaborate a little bit more

on some of your your your journey to

where you are and uh maybe diving into

we can dive into the specific Parts uh

when we get to them yep let's do it so

uh my journey starts uh in 2019 I

started at Texas A&M so gig maggies uh

started studying econ and that was also

completely random uh I I didn't really

know what I wanted to do in college but

I got fives on the AP exams for for

macro and micro so I was like this is

this is the way to go let's just start

with econ so I started with that my

freshman year my sophomore year I was

very fortunate to to get an internship

at the Federal Reserve because that was

during Co and internships were very much

not not happening uh I actually applied

for the research role so that was kind

of my background I wanted to do econ

research and I accidentally got placed

into the analytics division the treasury

now I had almost zero coding at that

point I had done a little bit of R for

one of my classes I think but that role

very much was a data analytics R and

modeling every role so that was my very

first introduction to to time series

work very first introduction to to

python pandas all all that stuff and

that was kind of a very very big flip

for me right I I went from just doing

more research type stuff to oh I'm I'm

very very interested in kind of the

modeling aspect like the actual applied

pieces of things so that after my

sophomore year when I went back to

school I I had an opportunity right I

could have I could have stopped and I

could have transferred to computer

science at teex A&M started from scratch

and done that and who knows what would

have happened if I had done that but

because I was already kind of on a

scholarship program I had my degree plan

all mapped out I decided you know what

I'll I'll keep studying econ let me try

to take a couple more stats classes in

the background to get that the math

fundamentals but let me just kind of

figure out the coding and computer

science aspect myself so my my junior

year was was very much just I want to

stop you and say at that time what did

you think you wanted to do because

everyone especially the the high

Achievers you're when you're in college

it's for something right at that point

you're like okay after University I want

to do X and clearly as you start to

enjoy the modeling more and you start

saying I'm going to learn computer

science on my own more what is that in

the purpose for or is it really just

exploratory at that point I it was

really just exploratory I mean econ I I

wasn't really sure because I because I

just did it because I came out of high

school liking it i didn't really know

where that was going to take me in terms

of a job after school and I'm in back

half of college now you're a junior you

have to start thinking about my junior

year internship is hopefully the one

where I get the return offer from right

and I didn't know if I wanted to

continue with the more econ focused

track and and this was really just a way

for me to explore it and get more

familiar with it right like at the

Federal Reserve it was a very small

piece of it wasn't even really computer

science right it was just like I'm I'm

writing code to model data so no data

structures no algorithms it was very

much just like plug and chug numbers get

something out and Tinker with it so I

really just enjoyed the the tinkering

with it and computer science kind of

learning those tools was like a way for

me to do that in my field so it it was

really fully exploratory I had no idea

where where I wanted to go with it

honestly yeah okay so you you decide not

to switch into computer science and

you're You're Now sort of are you uh

because a lot of people have this

question which is I I think I want to

learn coding but I don't know where to

start and obviously my answer is tends

to be like well just get your hands

dirty with anything try and solve a

problem that you have in your life or

whether it's for your homework or

whatever but use code to do that and

especially now with the the AI related

tools that's a lot easier but I'm

wondering um how did you sort of learn

on your own as you put it yeah yeah and

I think the way I did this I think has

helped me out a lot but I was very much

let's just do a project around something

like yes I mean there's a lot of courses

and there's a lot of really good YouTube

videos out there about like here's how

you do like take your CS 101 online

learn like your python or your

JavaScript or something and then maybe

take data structures and algorithms and

then try to do something else but but

for me I just wanted to to do something

with it right it for me it didn't make

sense to learn everything from scratch

because at that point I didn't want to

really do become a computer scientist or

become a software engineer I just wanted

to do something with it right so for for

me it was all about okay the most

approachable language is python it still

is let me just see if I can build

something with it right so I think my

very very first project was actually

around Spotify right love love music and

Spotify has like a very very

approachable API so you can get a bunch

of data right so I have a ton of

playlists I was just like can I just

grab all of my songs and visualize what

makes up one of my playlists and that in

my opinion that was the best way to get

started because right off the bat I hit

some bugs and then I solved some bugs

and then then the next line I hit

another set of bugs and that just kept

on happening until I got my very first

Jupiter notebook with with all of my all

of my songs and all of my data out there

but doing it that way I thought was was

very very helpful because it wasn't like

homework you submit something you get a

grade back from it right it was very

much I want to get to a goal I don't

know how I'm going to get to this goal I

know I can via Python and Via this

jupyter notebook and Via stack Overflow

CU chat jbt wasn't the thing at that

time right right so just doing it that

way I thought was was more helpful than

than taking a more traditional

traditionalist path yeah I definitely

think that as um you know code becomes

more for lack of a better term like

commoditized I think that just getting

your your hands dirty with it is really

the best way because courses are uh you

know not necessarily the information in

them doesn't become obsolete but the

structure is no longer as important um

and so I I I really like that attitude

and how you went project first I think

that's what I would recommend to a lot

of I will say though like so along with

the project first I I I do like I agree

those classes are fundamental and they

are very important right especially like

your data structures class right so

while I was at A&M I I did make it a

point because I had some free periods to

just see if I could go sit in on some

pieces of the data structures classes

right I think I think people forget that

in college you can you don't have to

give like your ID to hop into a class

there's usually open seats right and so

for a section or for for a piece that I

thought was interesting I would just

take that period And I would just go sit

in the class right and you also and Tas

are also students so you can always ask

a TA to give you some materials they'll

probably be like why are you asking me

for this but you can just do that right

you you don't need the the the test

period and the quiz period to learn it

if you just want to learn it you you you

can so I I I think that that alone um

sets you apart from many many people

because I think that most people don't

even want to go to the classes in their

degree and you're suggesting essentially

going to classes for a degree that

you're not even studying which I think I

mean I think that's still really cool

which is like if there's if you want to

learn something there are ways to go out

there and get it um so back back to

circling back on the story you know this

is Junior and then probably going into

senior year um and you graduate with an

econ degree what

next yeah that's good question so I I I

I was able to get the junior internship

as as like a a QA engineer didn't get

the return offer which which made sense

that was like my very first gig I didn't

even walk in knowing like Docker or any

of these things and I was expected to

know that it's it's all good um right so

actually we'll go back to like the start

of senior year right so right around

then was when I started kind of getting

into to blockchain a little bit right I

want to say to everyone this wasn't

crypto like specifically not crypto this

was blockchain from a very technical

perspective which I think is it's a very

important differentiator these days so

that was all because of one of my very

good friends Anish who also works with

us he kind of introduced me to the space

introduced me to his friends that went

to UT that were a part of the blockchain

club there and that was also for me

another way to just approach computer

science from a different angle right I

had I've been approaching CS from

machine learning data analysis and

modeling here was a way of approaching

computer science from a brand new

totally new teex stack totally New Angle

via blockchain right so towards the end

towards the end of my junior year summer

I actually decided to to start the

blockchain club ATM from scratch I

mean it's a bit more of like a high

level thing but I was like I I'm going

to text C&M I I'm just a student here I

I would like to leave my mark in some

shape or form and and leave something

behind starting a club was probably the

easiest and the best way to do it

because it's it's still alive today but

getting getting into blockchain tech

there learning some of those languages

also um like the tangent introduced me

to startups so the blockchain space

there's it's a it's it's obviously a

brand new space there's a lot of new

stuff going on which means there's a lot

of startups happening all the time and a

lot of that startup scene actually was

at UT Austin right so so being at T&M

I'm only an hour and a half away my best

friend goes to UT Austin my girlfriend

goes there so I'm I'm at UT Austin

pretty often and I'm meeting a bunch of

people

who CS Majors but also other random

Majors who are involved in this space

and they are doing some sort of startup

so Anish Anish and I like at the time my

in my senior year we were like hey what

if we're interested in in blockchain

we're specifically interested in zero

knowledge cryptography we we can get

into that if you want here but we were

like can we can we do a startup around

there right this is the the second

semester of my senior year I don't have

a job offer I'm really thinking about

grad school here I'm sending

applications but again I I don't really

know what where I'm going so in e i at

that point decided to kind of full send

the startup space I didn't graduate

Texas with a job lined up I with I had

grad school lined up but that's where

that was yeah just how can we get more

into the startup space how can we

explore this a bit more I I think that's

really interesting um and we can expand

upon this later maybe but the I think so

many people startups have become cool

right and so it's like doing a startup

has become cool high signal and so a lot

of people start with I want to build a

billion- dollar company rather than

interest in a particular space when

whereas you basically said that you were

introduced into entrepreneurship and

into startups through the venue of uh

blockchain but I think for many people

and I think actually most of the best

Founders that is what it is it's an

obsession over a problem and then Rec

recognizing the the sort of business

applications of the problem you just

obsessed over rather than starting with

I want you know to be a Founder for

example because again it's cool nowadays

um so I think that's really interesting

um so you you you graduate um are you

are you working on the startup in the

summer before grad school y y so we we

started working on it the that January

so this is what is this 2022 January I

think right yeah my senior year 2022

January we're we're we're working on it

putting as many hours as I can while I'm

in undergrad so it was a digital

identity idea basically we were like oh

can you can you prove pieces of your

driver's license without having to to

show your driver's license to someone

else right pretty pretty typical

application there so yeah we're we're

working and during that summer we

actually did like a an accelerator as

well it wasn't wasn't YC or anything but

it was basically a bunch of people came

in helped us with here's how you do

marketing here's how you do growth

here's some like smmes in the space and

we we worked all summer on that I had I

had an internship but it was like I was

putting 30 40 hours a week in the

internship but like after work we were

just going all in with Nish and Tom on

on the startup so yeah that's really

cool so I I think for a lot of people as

well um they look at you know obviously

I was I was a part of brev and I you

know reconnected with Nat who was who

was my childhood Arabic tutor um but I

think you also have like a cool story

but one that is not atypical which is

that you you started this company with

Anish who how long have you known Anish

yeah probably Middle School like sixth

seventh grade we've known each other

yeah and so so a lot of people would

think oh you're so lucky or they'll ask

the question how do you find a

co-founder which is interesting because

in this case you know your your

co-founder was was next to you you grew

with them which some would say is luck

but a lot of that is just keeping up

with relationships and keeping up um

growing together and then you know

Finding Tom that wasn't luck right that

was very deliberate that that was yeah

yeah what I will say about I mean I I

wouldn't be sitting here if it wasn't

for a niche I mean like I when I was at&

M right like I mean there there wasn't a

lot of Entrepreneurship there there was

some startup stuff there but if it

hadn't been for Anish just like really

almost like throwing me into this circle

introducing me to all these like

different people and these different

thoughts I like I I definitely don't

think I'd be doing a startup or any of

that here so so shout out Anish whenever

he sees this but yes meeting Tom that's

a that's a story right there so so in

that January when when Anish and I were

working on digital identity I was also a

part of this like online I think it's

like moo right massive online course

around cryptography because I was like

how am I going to learn this stuff I I

have to upskill basically so we're I'm

in this course it's it's huge there's

like a thousand something people in this

course right on telegram Tom Tom mfor

just hits me up and says hey I see

you're also in this we're in a couple

other telegram group chats cuz he was a

a part of the Syracuse uh blockchain

Club he's like hey do you want to try to

work through this course together and I

was like yeah as much time as I can

we'll do it we really did not end up

doing any of that at all it we it we

kind of lost contact there but in March

I joined another thing that was a part

of UT atin it was like another not

accelerator but more just like hey if

you're working on something cool we'll

Define some hours and we'll bring some

people into help Tom also joins that and

is that like pure chance or is this pure

chance this is pure chance literally

there's yeah pure chance so Anish and I

on one hand are working on digital

identity for the startup andish and I

are in this in this program Tom and I

also are interested in cryptography in

parallel we also end up working on some

sort of digital identity project in my

head I'm like wait I'm I'm doing this

with Tom I'm doing the same thing with

the niche can we just join together so

so in March of of 2022 I basically tell

Tom hey Tom meet my buddy in Niche we

all hop on a call and we're like hey

let's just join forces here for a a

little bit let's see where this goes

let's Tinker with stuff and let's see

what we can do right we do that things

are kind of going well it's it's diff

it's difficult problem still is but in

like may we learned that Tom's actually

going to be interning in Austin as well

andai are both in Austin Tom's like hey

I'm going to be living in west campus

and we're like wait so you're like a

three minute walk from us as well so

again this is pure chance not none of

this is planned and be I because we all

ended up working in Austin that's

that's the reason we we stuck together

because obviously if Tom had interned

somewhere else he was at AWS which is

crazy long hours and if if he if we're

not in person it it would have been much

more difficult to continue continue

forward yeah definitely so um I don't

want to I don't want to skip too far

ahead but you Anish and Tom um founded a

a a follow-up company um from the from

the cryptography um called Agora Labs Y

and um to to spoil that story which we

can get into to but agor Labs was

eventually acquired by brev which is the

startup that I was a part of that you

became a part of that then got acquired

by Nvidia and that's how we're both at

Nvidia um however you know you Anish and

Tom still live together you were telling

me about how you're going to your new

place um so I think it's just a really

really cool story of three college kids

again deciding you know what like what

Mark can we make like how can we develop

a solution to a problem and then Mark

get that that fundamentally is what

startups are and if you try and do it

any other way then you're bastardizing

it right don't be a solution looking for

a problem find the problem find the

people you want to work on it with and

then build a solution together and I

think that's that's really really cool

um so I don't know if there's more in

there that you want to you want to give

but I do think that your role at brev

was particularly interesting um and so

that's that's when I want to expand on a

little bit more because you know you we

talked about you starting from uh

economics and then U becoming more

interested in computer science and doing

stuff and then getting into like zero

shot cryptography and stuff like that

but then somewhere in there you became

technical enough and particularly in

machine learning that when you came to

brev pretty quickly I would say you you

earned the title of sort of like head of

ml or the head of machine learning at

brev and you were responsible for

dictating a lot of not only not only our

product but our educational materials so

where did that happen where did that

interest start and how did you develop

your your technical ability and machine

learning to be able to do that yeah yeah

that's it's a good question and I I I

would say I'll jump back to like the end

of that summer right so I mean solution

looking for a problem you just said it

that's kind of what the digital identity

problem was I

mean if you're going to go if you're

going to go show someone your ID at a

bar you're just going to show them and

then you're going to put it back in your

wallet and you don't really care too

much about what the bouncer sees you

really don't right so at the end day

when when there isn't as strong of like

a market need for it you're not going to

have customers right and we kept on

trying I mean we we worked we we talked

to as many people as possible we just

got to the point where like even

internally I was like I I care about the

idea a lot but I even myself I'm not too

worried about showing my ID right and if

I'm not like super privacy conscious

like that then I can't we can't be

following this idea right so luckily

around that time right LL 2 drops that

summer again we are not in the ml space

I mean I'm following it because I'm a

part of the community but I'm like oh

llms that's pretty dope right that

that's kind of cool but I was like let

me try my hand at at fine tuning llama

llama 27b when it drops why not right

very quickly I realized that the

Transformers and hugging face that the

stack wasn't evolved enough for me to

fine-tune it on a Google collab notebook

which is the the free version so I was

like I need to get a GPU from somewhere

now the question is where right I'm

looking around I'm not able to find one

on aw us can't find one any of the

clouds I find a GPU on this thing called

akashh Network right Akash is is a is a

blockchainbased cloud where you can

basically offer up your compute and you

get paid in tokens for it and other

people can rent out gpus from you right

so on one hand on Twitter I'm seeing I

can't get a GPU I don't have a100 I

don't have h100s and here I am using an

h100 to fine tune llama with crypto I

was like that's very interesting right

so I very quickly then we we decided to

drop the digital identity idea we were

like there's some potential here because

all of a sudden there is a market need

for these gpus we seem to have an

abundance of gpus here and there's no

way for people to use these gpus so

immediately we we dropped everything and

started building tooling around akos to

let people use it very easily and around

that time I was like let me I'm going to

focus in kind of go back to my roots

here with with time series and the more

machine learning aspect and learn as

much as I possibly can about this so I

around that time right uh summer ends I

go to grad school uh Columbia for a

semester even at Columbia right there's

a course path to take I don't really do

that course path I just cherry-pick some

classes because I'm like I really want

to take this class this class and this

class and those are the classes I take

so I really wouldn't have been able to

graduate if I had stayed because I

didn't take the right courses there

something that I want to add because um

people will criticize me um and

correctly say that you know like my

parents paid for my tuition at Georgia

Tech um which they did and I'm very

grateful um but a lot of people would

hear that what you just said about like

oh I was going to colia blah blah blah

for grad school and they're like oh like

money bags you know Bros all that no no

yeah exactly so so so to be able to like

I think that that's important to to

rationalize you know like going to grad

school and I want you to explain a

little bit more about like what you

wanted to get out of it you know yeah I

mean I I again self-learned a bunch of

computer science and undergrad but at

some point I realized that like if I

don't want to just be a software

engineer like just kind of like doing

the routine stuff I I I would like to do

a bit more than that and for that say

what you will you do need to have Rock

Solid fundamentals like you just you

just need to know some stuff and you

need to know it very well honestly yes I

do a lot of self-learning but I honestly

do do my best learning in a class so

Colombia was a way for me to was a way

for me to do a couple things actually it

was a way for me to get out of Texas for

a little bit right I had been in Texas

my whole life I went to school two hours

away from home it let me explore the

world live alone for a little bit pay

rent take on student loans and do stuff

that like real adults do and it let me

kind of explore New York for a little

bit and I was able to to Really benefit

from kind of the student Network and

Alumni network that Columbia had right

so all of those things in my head were

were very beneficial I I I still

personally to this day think that if you

do computer science in your undergrad

all four years unless you want to go

into a research track or a PhD track you

don't need a masters in computer science

I I really don't think you need it it it

costs a lot of money as well and I

really don't think it's it's worth it I

mean I was just taking classes right I

did I went to Columbia for very

different reasons I didn't go for

classes right I wanted to cherry-pick my

way around it I wanted to see if I can

talk to VC's in New York for Agora I

wanted to see if I can meet alumni who

might be able to invest in Agora and and

hear me out right versus just going and

and taking more classes

so I I think that's really important

because I think a lot of people es

especially on my platform wonder what

the relationship between education

either undergrad or grad school is with

entrepreneurship and actually think that

they're antagonistic with one another

they will be like oh yeah school is

great if you want to become an employee

but I want to be an entrepreneur and it

it sounds like you actually were

intentional in the school that you

picked to make the thing that you

eventually wanted to build have a higher

chance of success yes absolutely I mean

yeah they're not antagonistic at all I

mean like sure there are plenty of

Founders out there that have that have

unicorn companies that dropped out of

school and they and yes that's fine but

in the grand scheme of things what

percentage of people can do

that yeah it's possible but are you do

you really want to bank on like the 0o

point I don't know 000000 1% chance or

do you want to arm yourself and equip

yourself with as many tools as possible

so that when you're doing the startup

things get hard you need to do something

else you have the right tools with you

so you can succeed if you don't go to

school or if you if you decide to

actively say I will not go to school and

I will just do this well more power to

you unless you have the craziest idea

great but I don't think they're

antagonistic I mean you learn you're not

going to be able to do your startup

effectively unless you know all

everything you need to know right I mean

I agree with you so yeah I think that's

really interesting um and then you you

were actually building it this is the

other thing which is like you can build

it while going to school which

you have to make sacrifices yes you do

but you can right it's it's it's cliche

but like how bad do you want it right

like do

you you're in New York right yes there's

infinite stuff to do in New York there

are infinite parties to go to it's all

very alluring right but again there we

had a goal and each time I had a goal we

were like we are we are going to stay in

school because right now we want to play

it safe a little bit I want to continue

learning and we'll make the we'll make

the leap when we want to but for now

let's just kind of put our heads down

and just do the hard work right and we

fumbled a little bit I mean I'm not

going to lie we did at some point for

like a month and a half get super

obsessed like oh we could raise a preed

because we have some traction let's get

some money in the bank account then very

quickly we were like what are we going

to do with the money what's sure we

could raise a preed and then then do

what it doesn't matter we don't really

have that many customers we don't have

that many users we don't need to scale

up so yes you can you can do it in

school it's not you should do it in

school to start you yeah I I think that

that's great advice um and so you you

build aora and um when brev first found

out about Agora is a is a funny story um

we can go into how much depth you want

but um I think that fundamentally what's

important to note is that Agora was

similar to brev in terms of the problems

that we were trying to solve um we we

were solving them in actually even a

similar way it was just different um you

know compute sources and our backgrounds

were different and um ultimately ended

up essentially joining forces yeah yeah

so Anish Tom and I always knew that like

in order to really really find the

people that were going to use our

product because at the end of the day we

were making gpus easier to access well a

lot of people that need that are based

in San Francisco that's fact of the

matter I mean it's silic Valley right

this is the the heart of it all right

and so our goal really was how can we

get our El to SF as fast as possible

right so I was going to finish off my

first year of school and each Tom and I

were going to do the whole find some

small house here build an SF go to all

the events and just just work in SF to

really see like again you need

validation as quick as possible or else

you should move on or do something else

but you need that validation you need

the customers to tell you hey I don't

like this hey I like this we were

getting early signs of oh this is cool

it's super cheap it's very easy to use I

can use this but we needed to touch a

tap into a wider range of people right

so come February we're we're getting

through the semester I'm like we're

going to go to SF this summer it's going

to be great and then Tom Tom makes an

account on Bev's website brev at the

time was our biggest competitor in fact

the name of our group chat was brev Ops

right we we were Brad was our biggest

competitor Tom makes the account with

his agor Labs email with the company

email and what ends up happening is Nat

the the CEO of breev ends up reaching

out to Tom directly saying hey agor Labs

is cool you guys are doing cool stuff do

you guys want to meet now obviously a

Nishan on call like Tom what did you

just do why did you do it we this could

be the end right but but to Nat's credit

we we hopped on a call with Nat uh and

Alec

and I mean just from the first call

right the culture fit was right The

Vibes were aligned and Nat basically

after a couple calls was like

hey we'll use aor like we'll use your

your product that you guys have actually

built for breev itself so like give us

an API and we'll we'll use we'll use

we'll use your integration with AOS

Network right Tom and I worked for like

four nights straight just building the

API out we're like let's see if these

guys are going to use it I things worked

out on your end you probably know a

little bit more about that and it got to

the point where Nat's basically like hey

we are both building very very cool

stuff we are trying to solve the same

problem we both care very heavily about

making gpus easy to use for everyone and

not just these massive a companies we

both want that so what if we just join

forces do it together we'll bring you

guys out to SF and we can basically like

just achieve the dream together right

and I mean we were sold it it was it was

right there we were sold yeah so we went

from you know I think we were four

people at the time um then we we brought

on Anish Sean and Tom which was uh

really I mean 2024 was a crazy year for

brev in general and for all of us um but

with I can say with utter confidence

that we could not have been acquired by

Nvidia and done the things we needed to

do if it weren't for the acquisition of

the the Agora guys as we Cally refer to

you guys as and um I think that you

again were we so pivotal in that because

we uh needed someone who really

understood um what ml developers were

doing at the time and where they were

going because that's who our consumers

were and so you within brev were

essentially the the model ICP that could

then also go build um educational

resources which then I would make

YouTube videos on we had a nice little

Dynamic there um to then that was one of

our ways that we drove we drove uh user

acquisition which I think was was really

really cool um and yeah then we you know

grew brev and were were acquired by

Nvidia in July and that brings us you

know now we've been at Nvidia for eight

months now you're doing some really

really cool stuff on inference there um

but that's the the storyline of how you

can go from non-technical to being

acquired twice in about two years and

then we even you know you your

girlfriend ended up moving out here

which is great and um ended up in

introducing me to my girlfriend now um

so that's what I mean it's just crazy to

think that you know yeah we met probably

almost a year ago today and just how

crazy that gone um so first before I

forget and we'll put this at the

beginning as well um a lot of people are

going to want to uh follow along you

your advice um just see where you're

headed uh do you have any platforms or

anything that you want people to to know

about yeah Twitter Twitter's the the

best one I yeah throw it I'll throw it

wherever in like we'll throw we'll throw

his Twitter down in below make sure you

go follow him he's going to be I mean he

posts already some cool like ml stuff

but um yeah mostly mostly around ml

stuff some startup stuff I write a

little bit as well they're technical but

hopefully approachable to to everyone

yeah awesome so then now I want to dive

more into um sort of like potential

advice for those people who are maybe

you know wanting to get into a technical

field that's one of the questions I get

all the time which is like I'm

non-technical I want to become technical

now let's say that they're not at a

school and maybe they're not even as

driven or they're they feel like they

don't have enough time to go sit in on

other computer science classes right

what advice would you give to somebody

who studied a non-technical degree and

is like you know what this stuff

actually really does interest me I'm

committed like give me something that I

can do what what tangible advice would

you give that

person so if you're if you're committed

that's that's the biggest part because

that's the caveat on all of it because

if you're not if you're not committed

then no matter what advice what coures

what things you do it's not going to

work you you you and this is the

recurring theme right it's like you you

have to just willing to like pay your

dues and put in the work and that work

is not easy it's actually very difficult

it's annoying it's going to suck and at

times it's going to take away from other

vets of your life that you might not

want to happen right like it it's you

have to be willing to put to to

compromise but if you are if you are

driven right like the the first thing I

would say is like take a second and like

try to like map out what you'd like to

do in in this field right I think I

think a lot of

people's it's it's two very people

either I want to go Fang I want I want

to go work at meta I want to go work at

Apple Netflix those places or I want to

do my own startup it's like it's like a

bip forcation and I'm like there's a lot

99% of companies fall in the spectrum

between those two you you don't need to

go towards either extreme right so first

of all I would just kind of try to

figure out where you'd want to do like

what you want to do and even nowadays

within computer science it's not just CS

I mean there is at this point I feel

like it's it's CS applied to other

fields right if you want to just if you

want to just do plan software

engineering of course there's a lot of

roles for you but those roles are

probably a bit harder to find now

because there's so many people that are

wanting to just do that right I think

you're touching on something which I

think is really important for the for

people to sort of understand this is why

I think Georgia Tech where I attended

did it so great which is that every

degree no matter what you took you took

intro to CS nice wow that's pretty good

yeah and it was Python and it was great

and the reasoning why is because um and

I've talked about this concept before

but in order to become truly valuable

you could become the top .01% of

whatever you do right you're going to be

incredibly rare you're going to be in

incredibly valuable but by definition

that's going to be really hard to do

yeah what you also could do is become

really good at a couple things and then

find the intersection and all of a

sudden you're just as rare but it's a

lot easier to execute on and so if

you're a biology student who now is a

beast at py

game over I mean you're a killer you can

go work at any biotech company you'd

like I mean like there's a there there

it's it's funny and this is a fact that

I've heard from other people there is

like a there is a a significant need for

good computer scientists good

programmers at these other companies

people that do bio are incredible

researchers but again they don't have

like the the large scale systems the

ability to do something like that so if

if you have a bio background you've done

bio in the past

and you would like to add on computer

science don't just be like I'm going to

drop bio and I'm going to do computer

science and I'm going to go down the

software engineering rabbit hole no no

no like see where you can find an

overlap there right like so my very

first project for example like I said

was like something to do with Spotify

because I had not I didn't really know

what else to do you could do all I don't

even know what what's going on in the

biospace really but like you there's all

kinds of fun tangential where you could

try to apply bio to make your bio or

apply Python and CS to make your bio

life easier right that's and it'll also

probably be a bit more fun than just

dropping what you do right like if

you're kind of good at something you

have some background knowledge leverage

it as much as you possibly can and then

move forward with that right I yeah I

completely agree and you know bio is

still STEM related but I think it goes

it's true for PR it's true for

journalism it's true yeah I'm using bio

as an example but literally any other

field no matter technical non-technical

no matter what you're doing you could

always try to apply something to make

your life easier and I think in in the

age that we're currently in with AI

right with all that I think there is no

better and no easier time to do it right

I mean there's stories on Twitter online

everywhere you could have no coding

experience you could have like you have

never touched an IDE or Python and you

do something totally different but these

days you can literally open up chat GPT

claw and you can ask it hey I I do these

things I have these kind of

problems can I automate this or can I

make my life a little bit easier and you

will literally get the code for it

probably in your first second try right

and then just just from getting the code

output now you're going to have to

figure out I just got code how do I run

this what is it do I download and I like

what does it mean for me to run this

python code and just going down that

rabbit hole will teach you so much so

you're touching on something that I want

to expand on further which I think is

really important which is in the age of

AI tools uh I think a lot of people

think that they are

I I've heard the term either like

cheating or not learning when they're

when they're using these tools right

when in reality I think that the best

thing to do is like embrace them but

still learn and so I'm curious how you

would go about that right if you were

becoming technical with all the tools

many people would be like oh I don't

need to learn anymore I think that you

know you as being on the the back side

of it would be like okay what I did to

learn now is probably way

easier way easier I I'll give a tangible

example right so these past uh this past

week and a half I've had to learn a new

programming language rust right I I have

never used rust before I have never even

been in that ecosystem and I've had to

learn it right at at work we just got

approved for cursor Enterprise so now

I'm like I can use all that cursor has

to offer I can use the chat features and

all that right I still need to learn it

I can't just have cursor write all the

code for me and basically be like oh I

wrote all this right so the balance that

I struck there was this I have like our

company's rust code base in cursor what

I'm going to do is I'm going to still

throw bits and pieces of that code into

like the AI to ask it questions about it

but I'm going to turn off the tab

autocomplete so what that's doing is I

don't know what this code means I'm able

to learn faster with AI I'm able to

figure out how to do this with AI but

I'm still forced to to write my own code

I think that's that's like the balance

of it right like yes I I think if you're

if you have a crazy idea and you want to

just like execute on it turn all those

features on turn your tab autocomplete

on it and just get as much code out

there as possible and just try it but if

you're trying to like learn something

brand new right use it as a as like an

assistant to learn but not as like a

crutch right figure out how how you can

efficiently learn what you need to do

but you don't need the boiler plate

anymore you don't need it it's simple as

that we're not it's like oh like there's

calculators out now right oh you still

need to know how to do the math but no

one needs to do like long division by

themselves on hand anymore just pull out

your calculator and get the answer right

I I think that's a yeah that's a great

way of putting it um even when you

compare it to software engineering 15

years ago it's unrecognizable right and

so I don't know why people are acting

like you know this is the time where

where things change and either they

become automated or um you know it's

somehow adopting these new tools makes

you less of an engineer when again we

used to to write I say we but I actually

did in college but write in assembly and

it's like thank God we don't have to

write in assembly anymore we have

created further and further abstractions

to get that closer and closer to the

metal away from us yeah and this is just

the next stage of that and so yeah if we

don't have to write you know the the

crud operations for an API that we know

exactly what it should do we know why it

should do it the way it should do it but

we don't have to write those functions

anymore that's a good thing that's not a

bad thing yeah absolutely

um so that's that's great advice on

adopting AI um you are one of the

probably very few people who would come

on a podcast and could actually talk

about the space as it is today and where

you think it's going because ultimately

if you can at a high level explain what

you do at Nvidia now and then um some of

the most interesting things that you're

thinking about that people because

everyone sees the headlines with AI now

but deep into it so I want I want to

hear like your thoughts on the space

right now but first explain what you do

yeah so at Nvidia I work on our deep

learning algorithms team specifically on

the inference side so basically at a

high level I I'll give a little bit of

background right so when you think about

talking to Chad GPT or asking Chad GPT a

question what you do is you give it a

bunch of words and then you get a bunch

of words out right so in the background

what's happening is your words get

tokenized they get embedded they get

passed through a model and you get words

out one by one by one by one right so

like when you ask chat GPT a question

usually it's it's word word word word

word word word word word so my whole job

is basically making sure that you get

those words out much faster from the

model right super high level so

basically what that means is I'm I'm

working on optimizing model inference so

whether that means looking at new model

architectures and seeing if these might

help the model go faster whether that

means finding out new strategies on how

can I take this model and put it across

five or gpus and basically make the

model generate tokens even faster that's

kind of that's what I work on at a super

high level happy to dive into it more

but yeah no I think that makes a lot of

sense so um a tangible example right is

people have probably intrinsically known

that um for example I I've been playing

around with Gemini 2.0 flash which just

came out and it's significantly faster

than say chat GPT 40 yes and it's like

when I'm playing with it it just feels

like snappier right and so that that

feels good and you are essentially in

exactly that business right literally

exactly that yeah right so funny say the

Gemini flash right those are like these

new thinking models right like so like

deep seek R1 that's like super big on

Twitter for some reason is also one of

those models where you can ask it a

question but it'll give you a lot of

words back the whole thinking aspect

right that that's pretty computationally

intensive so so one of my big portions

of my job is figuring out how can we

make sure that these thinking models

when they're outputting a ton of tokens

they'll still be able to Output that

many tokens that fast as you get deeper

and deeper into the conversation right

because you're you're basically every

time you have a new you ask a new

question in the same like chat all of

those words are being thrown right back

into the model and you're getting a

bunch of output out and you do that five

or six times well the amount of tokens

that you're throwing back into the model

is there's a lot right so one of my big

L one my big big missions is how can we

make sure that when you're 10 11 12 30

chats deep you're still getting that

many tokens out that fast yeah that yeah

it's it's a really really actually hard

problem um and so yeah going back to to

my initial question because there's so

many players in the space now um and we

recently had both you know the the sort

of Chain of Thought thinking models um

you know probably starting with the big

one that everyone knows 01 um and they

didn't invent it but that was like the

when it became into the mainstream um

and so and then going into like agents

and all these other things what are you

a most excited about in terms of where

the space is going and then if you were

to try and give like a prediction over

the next year like how will we be

interacting with and using AI by the end

of 2025 yeah good question so I

think I think when you think about

inference right I think inference is

going to be commoditized if it hasn't

already very very fast I mean the cost

of using these models is rapidly

dropping now I know chat GPT like $200 a

month now to use all these new things

right but like when you when you think

about all of these like new open source

models like the Deep seeks the llamas

right the cost of using these on

Alternate providers is like pennies like

if you want to use deep seek served by

like the Deep seek company in China you

can like it's it's like it's literally

pennies to use it right so at that point

you can think that models are only

getting smarter but they're getting

cheaper to use right so all of a sudden

agentic applications I know that's like

pretty Buzzy but agents H have been a

thing for a while now I mean it's it's

it's putting the model in a loop and

calling some functions right at the end

of the day that's what an agent is but I

really think that that this year we're

going to see a bunch of brand new

applications with agents that actually

work right one of the big things about

agents is that once you get further and

further down down the chain they kind of

start to falter you might forget some

stuff or the quality just gets worse and

worse right but with these new thinking

models with the fact that I can get way

further down into a chain and not pay as

much money I think we're going to really

see a boom in in that space and I think

there's going to be like applications

that we don't even know about yet that

are that are going to happen right so so

from a tangible perspective does that

mean and not not somebody who's

necessarily um you know developing these

things although hopefully some of my

audience is Technical and if you're

building an agent related company I

think that's really cool I actually

talked to a guy a couple nights ago at a

Jord Tech event who is building um so

this is just super interesting to think

about the the ramification of a whole

new space opening up and the

opportunities that arise from that um

but he is in the business of just making

snapshots for agent history basically so

that if an agent is doing something and

it went to a point where you don't want

you can reverse the all of the actions

of the agent yeah to to a certain point

I'm like that's actually that makes a

lot of sense but that's a really

interesting problem that now has emerged

from giving a computer basically access

to go perform actions um of autonomously

right that's what an agent is um but

yeah anyways back to my question um so

that's more for like people who are

building it but there was a the world

fundamentally changed when we started

talking to a computer that really talked

like a human back and could perform

certain things the chat gbt moment as a

lot of people talk to it um what is the

equivalent of me who now I want to make

my life easier right agents already

exist what am I going to be able to hand

off to to agents to then be able to just

trust the to go do it how can I get my

time back

yeah it's a good

question I

think I think it'll

be it's a good question and like I think

there there's so much stuff that we do

that that is starting to already move

that route right I think I'll give like

a bit of like a roundabout answer to I

think a lot of like things that we use

and we kind of take for granted are

going to start adopting these agen take

AI systems themselves right so like for

example let's think like instacart right

like right now I go to instacart I add

the same groceries that I always do into

my cart and I get them ordered to me

here right I think we're going to start

seeing a wave where maybe it starts

becoming a bit predictive so it's not

I'm not saying it's going to order them

for you and you're going to have

groceries on your doorstep every

Thursday but maybe what might happen is

it might kind of start tracking your

habits a bit a bit better and it might

start kind of upfront already asking you

that hey

been a while since you've got this do

you want to get this and then it'll

automatically go and order it right or

maybe even I want to cook this type of

food I want to start making this kind

kind of Cuisine I might be able to start

telling instaart that and they might

start kind of automatically adding stuff

to my cart or being or suggesting stuff

like that so I'm not saying like I'm

like losing all control here instacart's

going to do it all for me I think bit by

bit slowly these like kind of this like

intelligence is going to start being

added to our everyday apps right I don't

know

no that that makes a lot of sense I

think that like I honestly I don't think

that we're going to see a crazy like I

don't I don't I really personally don't

see like another massive moment where

it's like wow like this agent can kind

of do it all for me I just think it's

going to slowly and slowly start

becoming a part of our life in ways that

we don't even really see it I just think

we're going to start using apps the same

ones that we always use and those apps

might just be a bit smarter and a bit

just oh nice it just kind of knows to do

that it just kind of knows already that

I should be doing something like like

this so you're saying and because this

is where a lot of people I think

catastrophize you know um it's not going

to be able to uh you know go off and

perform really really complicated tasks

flawlessly and replace you know like 90%

of knowledge workers by the end of 2025

no no no no I I think I mean dude they

can't even write code for me right on

the first try sometimes I I don't I see

why people think that I mean like yes

you use these models and you're like

holy crap

that would have taken me like 10 days to

do and I just got it done for me in 10

minutes right the whole deep research

thing where it's like oh I it would have

taken me so long to research this topic

and write this analysis and then deep

research gave it to me in like 10 days

that's great but like think of the

knowledge workers that are now going to

have access to that technology how much

more knowledgeable are they going to be

right I think it's it's all about like

not it's not going to take over for you

you're not going to lose everything but

you're just now going to be able to be

like 10x more powerful right off the bat

like I just think like the playing field

for everyone across the board with these

tools is now like one step higher right

it I don't think it's it's not like Doom

right so the best like we got cursor

approved at work now every single one of

those Engineers is now like a 10x

engineer right off the bat playing field

so now we expect 10x level work right

off the bat right I I think that's a

great way to put it um so in the in the

at sort of purview of code I think it

makes a lot of sense um when I've heard

it explained like this which is that

that doesn't mean that you know it's

going to replace software Engineers by

any means it doesn't mean that um you

necessarily need a bunch less software

Engineers while you could look at it as

they're more efficient therefore we need

less what I've heard of it as is now the

the tech debt that wasn't going to get

done for another six months can be done

today or that feature which wasn't

really the priority but it was 10th on

the list but since we can burn through

the other nine more quickly maybe we can

build that feature and so I think that

we're just going to have and you know on

the Deep research side of things it

might be you know the analysis that we

wouldn't have been able to put this this

speculative analysis let's say that we

wouldn't have been able to put an

analyst on because it would have been

too expensive y now we can Y and so I

think that um the amount of intelligence

for lack of a better word uh that we'll

be able to produce but from just

becoming more leveraged as a species

will in my opinion create a lot more

wealth a lot more value and a lot more

Innovation yeah I wanted something you

said there about like oh like AI code

assistants are going to take over the

job of soft rers I think people love

kind of saying that right but it I'm I'm

going to try to connect it back to

something I said a little bit earlier

about like if you were trying to get

into computer science you were trying to

start learning how to

code don't just approach it from like

I'm going to go become a software

engineer and do just software

engineering related activities like yes

to everyone out there unfortunately AI

can spin up a database and an API for

you much faster than you can yourself

yes it already can do that game over

right but think a bit like think more

like if you're going to get into coding

you should just know that oh I have like

the the boilerplate scaffolding already

built I mean the the hardest part about

a new project sometimes just figuring

out which database to use spinning it up

connecting it to your front end and then

connect that to your back end that's

difficult honestly

well AI can already do that so if you're

trying to get into it and you're trying

to learn all these brand new things

think of a think of a a a more difficult

problem or think of I have this problem

but this can already be done can I can I

do something even more with it right

like it yeah you can you can now think a

bit higher and harder and think more

ambitiously because all the boiler

plates already done for you I I think

that's yeah I think that's the right way

of looking at it obviously I think a lot

of people will think we're biased which

means I actually want to get into a

couple other reservations that I think a

lot of people have about the space that

I'd be curious to hear your response on

um one of them that I get a lot is that

um AI is very energy demanding and

therefore it's you know economic or not

economically uh insensitive to the

environment it's a you know one chat GPT

query is 10x the amount of energy is a

Google search you know that type of type

of analysis what what what are your

retort to that yeah I I will say I'm I'm

definitely not the most like

knowledgeable about the energy

requirements but that's something I

should should learn about and look into

a bit more

but what I'll say there is I I hope that

this can be a moment for us to to look

into maybe safer

cleaner ways to do energy right like I

know all these big cloud companies have

been like trying to figure out nuclear

power nowadays like everyone's trying to

buy into that nuclear power is is very

very is better for the environment it's

pretty powerful so far like we haven't

really been approaching it from the

right angle because of things that have

happened in history but and I still

think like we're looking at nuclear

power now for the wrong reason it's like

oh let's make our AI faster like it

should be more of a hey this could be

better for the environment this could be

better for long-term health of the earth

maybe we should look into nuclear power

for that reason but the fact that we're

looking into it anyway is like a good

thing right I I think I think it's it's

it's better for us overall yeah I

definitely need a b more research and

reading into in energy requirements for

sure I I I mean my my short response was

something similar about nuclear I think

that that's really cool that this might

be the the reason why we do it even if

it's you know not it's so

capitalistically focused but still if it

happens that's still a net positive the

other thing I would say is that every

technological innovation is inefficient

at the beginning um but get it's we're

already seeing it becomes so much more

efficient over time and it's it's so

short shortsighted to to be like this

thing doesn't provide enough value yet

for the amount of energy it's producing

because if the end goal which we assume

is something to the effect of AGI and we

can produce essentially human

intelligence for electrons think about

the amount of electrons that a human

consumes to get to the point like of a a

PhD student which now if we can just

like spin up some you know way less

electrons to be able to do that I think

that actually that's going to be much

more energy efficient yeah never thought

about it that way it's a good way to put

it yeah yeah um something that I want to

get dive more into um that's not as ml

focused but probably a bit more broadly

app applicable is uh the concept of of

work life balance because I think this

is one that um many people would look at

us and and think we're a little bit

extreme um but at the same time like we

both have girlfriends we both hang out

with the boys we both go to the gym like

we do all these things and a lot of

people are like well how is that

possible and so I'm wondering uh a what

your sort of own rationale is on like

how much you decide to to work whether

that's you know when you were at school

and building the startup or now at

Nvidia or while we were you know

building brev um while still maintaining

those other parts of your life and not

literally putting every single second of

every day into it yeah I'll I'll

approach it from two two angles like I

think the first thing is when I decided

to first try to pick up computer science

there was like two things there there

was one there was a fear that oh I'm

going to college for something I'm now

doing something totally else I'm not

switching to it I need to get good at it

so that fear and honestly that was like

a big kind of propellent for me at least

to just put in a lot of time and hours

into it because I knew that like I was

like this it's the real world I'm I'm

I'm I'm either going to make it or I'm

not going to make it so that means I

have to put myself in the best position

possible to kind of make it right so

that that did kind of demand a lot of

hours a lot of time and it did require

sacrifices I think one thing that I I

forgot maybe it was something I read

somewhere or someone told me it that's

always kind of resonated with me is is

is ruthless prioritization right that's

that's kind of the the way I try to

approach almost everything right and

it's yes I I I want to do a lot I I want

to do a lot of work I want to be very

high performing I don't want to skip the

gym I don't want to skip eating healthy

I would like to have a good relationship

with my family and my girlfriend to do

all that it just requires me to cut out

all of like the rest of it right like

that means that like while I'm sitting

in front of a computer I I don't really

have the luxury to do something else I

need to time block and I need to work

here and then not work here right I

think I I I think I I very much very

very much focus on ruthless

prioritization as as a thing right

now I I I honestly would say and this is

one thing I really admire about you

actually is like you have a I think you

have a very healthy and and good work

life balance I think sometimes I I I

fall into a bit of the unhealthy work

life balance like I'm not perfect by any

means there I I skip the gym it's it's

unfortunate sometimes I I work for too

many hours and maybe sometimes I I I

overdo it right but it's it's difficult

but what I will say is if you want to

like rise to the Top If you want to be

not the 0.1% but if you want to be in

the top like 10% of what you do you you

have to pay your dues you just have to

put be willing to put the hours up up

front you have to be okay with that

you're not otherwise it's not going to

work right right now what we're doing

here is with with Nvidia and startups

and all this we're we're we're planting

the seeds now so that we can enjoy like

the shade of the tree when we're 30 when

we're 40 right I think it's it got a bit

big on Twitter but like your 20s in my

opinion are not to just do everything

like you you need to you need to be very

very willing to put your head down get

something out of it so that later on you

can enjoy it and you can look back and

say look I know how to do hard work I

have failed I have done very difficult

things and I know how to approach a very

difficult problem because I've done it

so many times earlier on in life right I

I think it's a great attitude um you

know I I'm someone who yeah I do um you

know make sure that I go to the gym and

I sleep enough and I do all that but I

also work really really hard oh yeah um

and I think that it was actually on a

podcast I really really like this

analogy it was by one of the co-founders

ERS of Netflix and he described um you

know like working hard in your 20s as um

as as part of an Iron Man you know you

you're doing like the swimming part of

the Iron Man and um everyone jumps in

the pool and it's like if you just try

and pace yourself at the beginning

you're getting kicked in the face

there's a ton of people like you're

you're just trying to you know get ahead

but if you just Sprint really hard and

get ahead of the pack then you're in the

open Waters and at that point you can

pace yourself you take a little bit more

closely and I think that that's how

really starting your career is when you

have no skills no experience no anything

it's like if you're prioritizing oh I

want to make sure that I stop start at

9:00 am stop at 5:00 pm exactly I don't

if I I want the flexibility to work

remote I want you know and if they ask

too much of me then I'm GNA quit or I'm

going to switch jobs then it's like you

will never build yourself up to the

point where you have leverage to make

the decisions about actually having

Freedom yeah yeah

yes absolutely like why like yeah that

that the mindset doesn't really kind of

make sense there right it's

like why like you are just starting off

you don't really have leverage at all

yet you you you can't battle like so you

have to be able to put yourself in the

best position so you can rise up and

kind of move forward unless you're okay

with staying stagnant which in that case

more power to you hopefully they're not

watching this yeah like then it's that

but the work from home type thing right

that's that's an interesting one right

it's like I I I feel like I do see a lot

of people complain about oh I'm having

to go into office oh like they're

forcing me to go in a couple days a week

I think the most I think the most

important part about brev and the reason

we were able to be successful was

because Nat and Alec and you guys had

the intensely non- remote literally on

the brev website it says we are

intensely non- remote in San Francisco

there was no work from home at brev all

of us were in that in like that room

every single day because there is there

is no substitute for it if I have a

problem that I want to solve it's a lot

easier for me to TAP you on the shoulder

than try to slack huddle you or be like

yo are you free for a call right you you

can't really get the the creative juices

flowing you can't really do as much if

you're not willing to go in in person

and one thing that Nat has said that I

love a lot is if you want to get lucky

you have to increase your surface area

for luck you want to talk to the VPS at

your company you're not going to talk to

them on slack you're going to have to

bump bump into them at some point or get

lucky at work so you get to talk to

someone right otherwise it's you're not

going to not going to make it really

100% I think that exactly what you're

were saying I mean we're both you are

you 24 24 24 I'm 25 and you know we have

the pleasure now of working working at

Nvidia where there are some really

really talented people and um I I tell

everyone if you know if you're if you're

doing your first job out of college it

100% should be in person because that's

how are you going to become friends with

the guy or the girl who um significant L

ahead of you in their career and is

going to either give you an opportunity

that you wouldn't have gotten otherwise

or just Mentor you um I'm like it's

going to happen probably at like a

company happy hour or like a social

event or just bumping into each other

often times at the office it's not going

to happen from me putting a weekly sync

on your calendar because they're gonna

say no and so I think that like you you

should absolutely go in person um

especially early on in your career if

you're you know what if you're in your

40s and you have a a wife and kids or a

husband and kids that you're taking care

of and they're young I understand

wanting to have a little bit more

flexibility but that's exactly what

we're talking about do the work now so

you have the flexibility L you can't

afford to not go into the office right

and also it's like lucky talking of VPS

that's all great but like I think in

school when we're in college you have

classmates you go to class with you have

homies you you live with them in your

dorm but like when you get to like the

real world and you're at a job it can

definitely get a little bit like oh like

I'm not around any of my friends I'm

around totally new people and I I don't

see every one when I wake up and go to

bed so going in person also just lets

you like still like be a bit social

right like like it's when I work from

home sometimes it's like hard I'm like

wow I haven't spoken a single word in

like 10 hours I would like to go talk to

someone just to talk right going in

office lets you still kind of build up

those friendships those connections that

other that you had you had for granted

in college but you definitely don't have

in the real world like yeah that no 100%

agree um speaking of college um because

I I think that you know we've

effectively uh talked about that topic I

want I want to Pivot a little bit

because a lot of people again will look

at um me and be like okay well you went

to Georgia Tech which is a a really

well-known school for computer science

or computer engineering granted not like

the mits of the world um but you you

went to to A&M which is still a great

great school but um you were you were

saying something a bit off camera which

is that um it still did not feel quote

unquote big enough in terms of the the

scale of opportunities that were

presented to you that you actually you

know you mentioned uh finding

opportunities at like UT Austin and

doing that stuff so I want to hear more

about like how you pushed through um

finding the the growing into what you

think you should have been um or as you

put it like the big fish in the small

pond uh type feeling yeah A&M is a

fantastic school I learned a lot while

being there and I loved I loved my whole

time while I was there but at the same

time A&M is not known for its economics

it's not the target program and A&M as a

whole is not a Target school it is for

like petroleum engineering and for some

of those disciplines but not for

business or not for for econ it's not

that's a fact right we can be around the

bush but that's what it is I didn't have

the best recruiters coming to Career

Fairs I didn't really have Career Fairs

for for econ right

so there I said all that that

sucks but what that means is all of a

sudden it's a lot easier to Stand Out by

doing a lot less so let's compare that

to the mits of the world everyone there

is super driven super motivated the

programs towards the top of the nation

you have the best professors coming in

teaching you so to stand out in front of

all those people takes an extraordinary

amount of work doable but it is

extraordinary standing out in front of a

professor or or or getting that extra

opportunity for research at A&M actually

wasn't that difficult because there

weren't as many people asking for it so

a professor wants to take on a couple

kids for research well there's only a

couple kids coming to her for research I

was one of those kids right so it was a

lot easier in that way to stand out

now that sounds easy and all right like

you have to be willing to take that

opportunity which is like the hardest

part it's all easier said than done you

have to be okay with okay I'm going to

go into this office hours even though I

don't need it just to chat with the

professor a bit more I have no questions

but I'm just going to do it just to

build up the Rapport a little bit right

here here an example is my my career

counselor or my like my adviser at at

A&M right I'm I'm saying I'm able to go

to these different classes and all that

right she was extremely helpful with me

being able to take classes at certain

times and also basically say even though

you're not allowed to exempt this class

because you've taken this stats class

here I'll have it count for this right

that's not because I was at a school

like A&M for economics that's possible

because there's not as many kids asking

for it and because yes the class is the

class is the same right it was a lot

easier to do that so if you go to one of

these like kind of non-target schools

I'll say and you you want to get a

little bit more and find more there's

opportunity everywhere because no one

else is grabbing it so you can you can

do it the the the UT Austin blockchain

Club has been around since I don't even

know like 2018 2019 it's like one of the

best ones in the nation right ANM didn't

even have that community so for me it

was like I can actually do some work and

spin this community up take some advice

from my UT friend

but build a whole club from scratch ATM

because it literally doesn't exist right

so there there's a lot yeah I think I

think that's great I mean it's such a

glass half empty or glass half full type

of mindset that um especially a lot of

people who want to take the Doomer

mentality will look at whatever

opportunities they could have had if

they had done X whether that's something

they a decision they made or even just

like a circumstance that they were born

under

um and uh you're you're absolutely

correct which is that you know no matter

what you're in there's a way to leverage

that to actually probably some kind of

Advantage but even acknowledging the

fact that there it might be harder to

get land certain opportunities which

means are you willing to to put in that

extra work um I think I reposted a video

today that was from a year ago but

people ask me like oh does your GPA and

the college you go-to matter and I'm

like yes of course it does but it's not

the only thing that matters and it's

like you can get a go beyond that but

like let's not fool ourselves here yeah

I there has to be a you can't do one of

two extremes you have to have a balance

in all of it yeah like yeah I agree I I

very much agree um yeah that's yeah it's

really interesting um something that is

also unique about you that I think a lot

of people would look up to um or maybe

would be like how is that possible is

that throughout all of this throughout

College throughout building a startup

throughout getting acquired um you've

had a girlfriend and you've had a

girlfriend who is also very driven also

very hardworking and at least from what

I can tell um you know you you've

developed a a very deep relationship you

guys respect each other a lot and you

guys both work very hard I think a lot

of people are curious um again how how

that happens how do you maintain that um

I so many people ask me like should I

have a relationship in college so I

guess we could start there but what's

your opinion on that

like with all things we've been saying

it's a balance right and I I'm I'm like

very fortunate that that my girlfriend

Olivia she is way smarter than me and

honestly way more driven than me I mean

she's she's insane but I think so she

went to UT Austin I went to Texas right

so so we did a bit of long distance

there I think that was the best possible

thing we could have done early on so

movie started dating five years ago

that's like when I was like 19 right

technically so that's pretty early to

have a long-term list ship but us not

being going to the same school what that

what I'm saying what that allowed us to

do is like grow grow together but also

grow separately right like it I was able

to take time and explore all of my own

interests and there were nights where

yeah I'm not going to be able to hang

out because like I'm working same with

her right going to two different schools

honestly I would say like helped with

that it made our relationship stronger

like it's now she goes to Stanford so

we're only 45 minutes apart so it's it's

better now but yeah I will say that it's

all balance right there have been plenty

of times I remember there was there was

a time that she came to New York while

when I was at Columbia and that was

literally the week where I had a midterm

but we had like a super big thing that

we wanted to deliver for Agora so she

came to visit that week and she

literally toured New York without me she

like did all the sightseeing without me

because I was like busy working

right she had to be okay with that and I

I I mean I made it very clear that hey I

have a lot of stuff to do this week I'm

really happy that you're here maybe I

could have played things a little bit

differently but like the fact of the

matter is now now you're here right like

I

think being in a relationship like that

kind of being able to grow like that and

being okay both people being okay they

like hey I know you have a lot going on

I have a lot going on I really

appreciate your company I really

appreciate you I'm so happy you're here

I'm going to be doing this you're going

to be doing this and being okay with

that I think I think is was very crucial

and is very important for for us yeah I

I think that's such great advice because

so many people especially young men will

uh I think it's honestly preached to

them that um somehow a relationship is a

distraction especially for the super

driven people and um my my response to

that is always like certain people could

be yeah but if you are if essentially

choosing the right person in a lot of

ways um they will be so bought into your

vision in the same way that you know

hopefully they have their own Vision

that they are never going to put

themselves in a in a position to

jeopardize you reaching your goals and

and and if you find that person and

actually in a lot of times it'll be uh

constructive to your goals it'll make it

easier if you have someone who's again

uh loving you and participating in the

when they need to to help you reach your

goals then I think that that's really

what it is about to have a relationship

and um it it makes it so much more worth

it because then so many people are like

oh you know it's the same guys who are

like oh I won't get a girlfriend until

I'm successful and then we'll say things

like oh but girls are gold diggers and

they only want guys for success and it's

like that you can't hold both those

beliefs at the same time yeah yeah two

extremes you cannot be on the extremes

you have to find have to find a have to

find a middle I I I talk about Nuance

all the time on this platform um so in

order to kind of like wrap up because I

think that uh you know this was a

phenomenal conversation uh my audience

really is the kind of like 18 to

24-year-old typically male I do have

some female audience but we'll we'll

preach to them real quick which is um I

think you like me is someone who they

probably look up to right and so um if

there was an 18-year-old man or 18

through 24 year old man who who feels

kind of lost right now who looks at

someone like you or me and says I want

to get where to where they are or be on

the path to where they are um but I I

generally feel kind of

loss what would be your advice and we'll

keep it fairly high level right without

being super prescriptive um I have an

idea of like what you might say cuz I

you know talk to these people every day

but I'm I'm curious what your advice to

that person is yeah yeah what I I want

to make sure I'm not being prescriptive

because everyone's everyone's on their

own path everyone's everyone's doing

their own thing right it's not a linear

for anyone I

would I would I would just say that that

when you maybe look at like like you and

I path what you see is like the LinkedIn

Journey you see here to here to here to

here right that's like that's what you

see and it's like wow it's super cool

but like what the LinkedIn or what these

platform like even this podcast doesn't

really show is like all of the super

small steps right it's there were a lot

of like just individual days where no it

did not feel great at all no this it was

hard and a lot of the days were like I'm

lost I don't really know what I'm doing

here yeah it's not a not a linear

Journey for everyone right like you you

have to be willing to be okay with like

the small steps it it's a day-by-day

process you're not going to you're not

going to get any overnight success

success unless you're a part of the 0.1%

and you can't Bank on that right like

during the Agora days during the time

where I was first learning computer

science that was a very very scary time

and I felt lost I am telling my mom hey

like yes I'm an econ major hey I'm not

going to switch but I'm going to spend

my days doing something totally else and

I'm about to try to enter a very

competitive field where there's a lot

going on there's people that are a lot

smarter than me but I'm going to try

right that I felt lost honestly I did

but

no matter how difficult and I know it's

again easier said than done but no

matter how difficult it is you have to

just day by day just keep pushing keep

keep chipping away chipping away

chipping away chipping away CH because

like what's the alternative right either

you either you keep chipping away and

you increase your surface area for luck

and you you you you learn a little bit

more as you go or you don't do that

right when when when you're in like the

18 to to 24ish range right like I think

I think you your only goal in my opinion

should be to fail upwards do something

you are probably going to fail because

you're only 18 to 24 but if the only

thing that came out of it is I learned

something new or I learned something

then it was a complete win that's it's

all you need I think it's great advice

uh a great piece to end on work hard uh

be humble about it and don't expect to

have all the answers and I think you

know for 18 to 24 that's that's really

the only advice I would give as well um

dude thanks so much for coming on um if

you don't already which most of you

probably don't go follow isan on Twitter

um he has so much advice to give beyond

what we did on this platform and he's

doing some really really cool stuff at

Nvidia definitely someone you'll want to

you'll want to follow along on as his

journey only goes as he continues to

fill upward from

here appreciate you man thank you so

much for having me