0:05

Welcome. To the Code to these are

0:07

cast powered by the Deaf community where we

0:09

talked to before their coding journey as of

0:12

helping you on yours. I'm your host Iran

0:14

and today we're talking about Ai with Rob

0:16

Philo cofounder and chief an officer at The

0:18

Story that. Oh I'm

0:20

living the dream. I literally for sleep with my

0:23

left a of it. I wake up three to

0:25

mourn the excited to see how my latest experiment

0:27

is going Everyday when I have a spare moment

0:29

I'm checking the latest news, checking how people are

0:31

liking the features on story graph see how I

0:33

can approve them. The everyday is a new challenge

0:35

everyday that almost almost like a new world on

0:38

a weekly basis with the new news that's coming

0:40

out. So yeah I can imagine it being any

0:42

better than us. That shares

0:44

how a trip and his friends Tesla let

0:46

him to enroll in his first A I

0:48

Coarse language models. He's created a long as

0:51

journey and where he recommends those need a

0:53

machine learning should start after this. Are

1:04

you looking to connect with the diverse

1:06

audience of developers? Look no further. You

1:08

can partner with us here at the

1:10

Code to Be podcast and will help

1:12

get your message out to are incredible

1:14

listeners in an app just like this

1:17

one led by me, your host contact

1:19

us by emailing sponsors at Cotonou be.org.

1:25

Thinks. Worshipping here! Hey thanks for having me!

1:28

So you are not only the cofounder and

1:30

chief a officer of the story got that?

1:32

You are also my husband and I'm so

1:34

excited to have you on the show because

1:36

this season we're focused entirely on a I

1:39

am trying to get more people who are

1:41

interested in a i figure out how to

1:43

break into the industry and we're trying to

1:45

figure out what guess we should have that

1:47

of hopefully had kind of that code new

1:50

be journey because traditionally only think about people

1:52

who do a I, we think of Ph

1:54

D's and you know start ups. With hundreds

1:56

of millions of dollars of funding have been

1:58

studying this at Mit and. Birds and

2:00

you've been path but you kind of broke

2:02

into a i later and your career kind

2:05

of as a newbie and I thought oh

2:07

my goodness I have a code new be

2:09

story that I'm married to his let's bring

2:11

him on A New Hampshire. Your story some super

2:13

pumped a happy on the show? Yeah me too. Thanks

2:15

for inviting! I've been a big fan of the So

2:17

have been listening since Absurd One. A you

2:20

actually were are very first unpaid set.

2:22

Aside as an editor at

2:24

Never Enough last. Us A see how

2:26

many years ago they gives a much for your

2:28

title is help All those years that you were

2:30

apart as editor and now you're back as I

2:32

guess he's upgraded said the for yeah. Of course

2:35

I feel like I was in the shark tank

2:37

as an entrepreneur and now I'm in a sort

2:39

think as a sorry am ready to get on

2:41

earth I love at All right? Let's do this.

2:44

So rumor has it that you first got

2:46

into tears when you are just two years

2:48

old. Is that true? It. Is true.

2:50

My dad put me in front of a computer

2:52

age to we had the I B M P

2:54

C junior think was called a note was a

2:56

missile was a Dos prompt know gooey yeah I

2:58

wonder what their computer looks like Now a kind

3:00

of want to sit in front of understood get

3:02

the nostalgic feeling if I see if I remember

3:05

anything because. I almost want to

3:07

take our daughter and put her in front of

3:09

a terminal right now. Was actually went mad at

3:11

aides who said about i've had a single differences

3:13

and have a good. What she's almost is is

3:15

twenty months. So we ever a couple more months the

3:17

middle put her in front of a current laptop. See

3:19

what happens to do you remember

3:21

having any feelings? Associated with technology at

3:24

that time. he was so young. But.

3:26

Sato. So cool! I was a big gamer

3:28

as a kid and thing Mario Brothers and

3:30

a member plane Doom the first time I

3:33

saw Three D graphics on a. Screen.

3:35

Of that do maxi qualifies as that. I

3:37

was, as is, kind of shocked that I'm

3:39

and Will from Stein Three D. So yeah,

3:41

I remember thinking from a young age that

3:44

this is something that has the potential. To

3:46

this be really awesome. Worth the very least to be

3:48

lot of fun and I knew that the path to

3:50

get there. Was. To being a

3:53

programmer. So as the at the youngest

3:55

age I can remember I wanted to

3:57

be a programmer making video games so

3:59

I knew. You know today, everyone

4:01

has a computer Every kid has of

4:03

phone or tablet or your has access

4:05

to technology is so ubiquitous he sees

4:07

the back than people generally didn't have

4:09

computers in their homes. I was kind

4:11

of abnormal, doubt a novelty and you

4:14

are actually accused of cheating when you

4:16

were younger because your teacher assume that

4:18

you could not possibly have that access

4:20

to a computer and done your homework

4:22

yourself tells a story. There. Was

4:24

quite ridiculous. Not, I think back on it. I

4:26

was spending the weekend with my dad and I

4:28

had a book report or some type of reports

4:31

that I had the right before the week on

4:33

is over. And he. Said

4:35

he robbed. We're going to focus on this. We're gonna

4:37

have a good weekend, but we're also going to do

4:39

homework done. So he drove me to the library, checked

4:41

out all the books that I would need, and I'd.

4:44

Sat. In front computer all weekend typing out

4:46

this report. Remember. Those Microsoft Word

4:48

or Word Pro First an aunt member exactly

4:50

what the ultimate processor was but a member

4:52

spending all weekend on it. And at the

4:55

end that a weekend I printed out on

4:57

Monday eyeball of the teacher handed it in

4:59

and scan a game. It is weird book.

5:02

And later on as he was grading papers

5:04

he called me over and said rob. Did

5:07

you get any help on this? And as

5:09

a kid, articles and second or third grade I just

5:11

put my head down and I'm thinking. Well

5:13

my dad brought me to the library

5:15

so I guess I had help and

5:17

I responded yes I had openness and

5:19

see Jamie the paperback. Kenya. Little

5:21

speeds on the value of doing your own work.

5:23

Told me to go home and do it over

5:25

myself this time, which in my view mental tag

5:28

as I can go to the library and I

5:30

did it by hand because I was it with

5:32

my dad. I didn't have access to the computer

5:34

for that week and when I brought the report

5:36

back in the next day. To give

5:38

me a long speech on Sale Good it is to

5:40

do work on your own And she's looking at the

5:43

paper that I used pen and paper to write out

5:45

and it looked more what's he expected from a second

5:47

grader in. An unknown nineteen ninety

5:49

you at every year that was and I'd still than

5:51

it really understand what was going on. But and due

5:53

to teach was happy so I was done. But when

5:55

I told my dad about that. I. remember

5:57

how angry he was and i remember the the

6:00

links he had of almost a

6:02

betrayal where he saw his son

6:05

work very hard at a task using the

6:07

latest technology possible and to be accused

6:09

of cheating without giving an explanation

6:11

of why. She never said

6:13

why she thought I was cheating. I just was

6:15

accused of it. I'm sorry

6:17

you went through that and I'm very excited

6:20

that you had the opportunity to have computers

6:22

at such a young age and was even

6:24

in a place where you were so good

6:26

that you could be accused of cheating is

6:28

kind of a cool place to be in. So you

6:30

grew up with technology introduced to you by

6:32

your father. Tell me about

6:35

how that love of technology

6:37

grew or maybe changed shape over time up

6:39

to the point where you go to high school.

6:42

So growing up I was always the tech kid of

6:44

my friend group. I was always the tech person to

6:46

talk to and all the family get together. I was

6:49

the one who solved problems and that's

6:51

kind of how I got started playing video

6:53

games on my computer just trying to figure

6:55

out how do you even install this thing.

6:57

I always wanted to do something more with

6:59

technology. I wanted to create the

7:01

best video game you've ever seen and the way to

7:03

do that I had to be a computer science major.

7:06

So that's why I focused on getting into a college

7:08

that had the best computer science degree

7:11

that I can get. But I remember in high

7:13

school despite being the tech guy, despite having a

7:15

reputation for knowing everything about

7:17

technology, I still got some flack

7:19

about it. Some of my friends who some

7:21

of them were also technical said, hey Rob,

7:24

you don't even know HTML. I know HTML. I'm

7:27

going to go to college. I'm going to learn how to

7:29

program because I know HTML but you don't even know HTML.

7:32

How do you expect to excel in

7:34

college if you don't even know

7:36

any programming languages? And I thought about that

7:38

and I was like maybe 16, 17 at

7:41

the time and I remember saying, yeah,

7:43

I don't know. That's why I want to go to college. That's

7:45

why I want to learn it. I'm not supposed to know before

7:47

I go. I'm supposed to know when I go. But it was

7:49

interesting to me that even at that young age

7:51

at 16 or 17, I was expected to

7:55

know things already. And when

7:57

I didn't know those things, it was like gatekeeping. The

8:00

keep away from my dream

8:02

of. Being. A video game programmer.

8:04

so it's it's kind of sad way to

8:06

that. Ended up. Tennis

8:09

and says that is interesting that

8:11

evidence as anyone to know organic

8:13

chemistry before they got to organic

8:15

chemistry. So things like a very.

8:17

Unrealistic. Playing for a kid to expect

8:19

you to know before you go into the

8:22

school when you know things are trained. The

8:25

you decided to be a computer science major

8:27

and you went to the Universe Email and

8:29

I believe which is where I went to

8:31

as well. And when you were there did

8:33

that computer science degree end up being everything

8:35

he hoped it would be? Unfortunately, it

8:37

was kind of a disaster for me.

8:39

He didn't work out at all. My

8:41

first. Two semesters with okay,

8:44

but as soon as I've gotten

8:46

to the more complicated, higher level

8:48

calculus classes, I started having

8:50

trouble and I remember at one point I.

8:53

Put. My whole dream on hold almost.

8:55

I member thinking is this really. What?

8:58

I want to do Yes! But. Is

9:00

this something that I can do? I

9:02

don't know and at that time I put

9:04

a message form post out on the

9:06

I Gn internet gaming network forums and I

9:09

asked I said he. I. Know there

9:11

are some programmers here. I. Want to be

9:13

a programmer? I want to make the against like your

9:15

but I'm not very good at math and I'm. I'm

9:18

about to maybe fill out of this

9:20

high level calculus class. Can.

9:22

I excel as a computer

9:24

programmer, as a game developer

9:26

without being really good, a

9:28

calculus without knowing the math,

9:30

and the response was unequivocally.

9:33

No. You're. Not going to be successful. You

9:35

need to be able to pass these classes. You need to

9:37

be able to understand the math. And or

9:39

to do this And. Again,

9:41

I felt like I being gatekeeper but this

9:44

time is that were real assault. More like

9:46

well I need a computer science degree. To

9:49

get this job. The. Computer science

9:51

degree has these requirements. I cannot that the

9:53

requirements therefore I guess is not for me

9:55

So actually switched majors my sophomore year and

9:58

went to economics because does the stuff. Reading

10:00

game I liked from a high school. Economic

10:02

class that I took in a city I guess

10:04

ago the economics but the very first chance I

10:06

got when I graduated my very first job, I

10:08

made sure that job was intact. So

10:11

now that you are into doesn't attack

10:13

for many many years in almost twenty

10:15

years I would say I'm not quite

10:17

twenty years and. His sister doesn't Five? Yes,

10:19

Yes, Almost twenty years where they write

10:21

to you need mathematics to do the

10:23

planet. Computer Science Programming says the you

10:25

wanted to do. No, no, not not

10:28

at all. Maybe if I am

10:30

inventing a new paradigm. Maybe if

10:32

I'm inventing. A lot

10:34

of some new algorithm I do. Need

10:37

math? The vast majority of

10:39

people don't need that. Is

10:41

it upset you now thinking about that advice that

10:44

you thought on looking back at that time and

10:46

knowing that you. Change your trajectory

10:48

because as. That information.

10:50

it really does and I wish that

10:52

may be. The advice I got was.

10:55

Well. To get a degree at this. University.

10:58

You need to be able to pass these classes,

11:00

but there are other avenues for you. You can

11:02

do X or y or z o E to

11:05

learn on your own or you could. Get.

11:07

A tutor or whatever else. But

11:09

those paths weren't visible to me.

11:11

They were presented to me. And

11:13

at no point. Out of my

11:15

advisors was I was on the track team.

11:18

I have access to economic advisors. Nobody

11:20

that was advising me, nobody that I taught

11:22

to. Presented. A path. To

11:25

being a programmer. Without. Advanced

11:27

Calculus. So. I quit.

11:30

Hard to see all the time. I

11:33

was devastated. It was my dream forever

11:35

and. To. Have it taken from me. Not.

11:38

Because of an external force does. Now I

11:40

saw it. I saw it as I. This

11:42

wasn't good enough Those people have. You will

11:44

suffer from the ocean. Plan. Oh

11:47

of my plan was. His. His graduate get

11:49

a degree in something that. I

11:51

would enjoy What's. The. Our joy? economics

11:53

as I enjoyed the stock trading in

11:55

the. High school class. I took. And.

11:57

I said. Well. My very first job.

12:00

Going to be intact and I made sure I

12:02

first job it had nothing to do with economics.

12:04

My first job out of college was. A

12:06

recruiter. So I was looking at. Hundreds.

12:09

And hundreds of tech resumes all

12:11

day. But. I was in a tech company

12:13

and was I had muscle in a door into I can Move

12:15

Around. See your goal was

12:17

maybe the job itself isn't setting a goal that

12:19

it's a reason the right industry and was your

12:22

plan to your help to kind of find your

12:24

way into a more technical role at some point

12:26

was that kind of the The strategy has a

12:28

point. Yes, Thousand as you. Say

12:31

you are the tech recruiter. How's it feel to

12:33

be on the other side of the table? You're

12:35

the one you know begging for jobs trying to

12:37

prove your worth, Get through that degree program and

12:39

now you're the one picking people how they feel.

12:42

Oh it was weird because I felt like I was

12:44

in the position. That other people

12:46

were in stopping me from become see

12:48

my dreams goes. The vast majority of

12:51

resumes I saw word declined and the.

12:54

Reasoning. I had or the marching orders

12:56

I had my top were the need to

12:58

have at least at the voice I Gps

13:00

and indeed have come from a prestigious university.

13:03

So. You could have been the best candidate

13:05

ever and as had trouble in one or two

13:07

other classes didn't have the people who buys and.

13:10

Greed and talk to you. And

13:12

now that again he than a professional in the

13:14

industry for so many years and you've done the

13:17

job done. The work is done, the hiring back

13:19

in your. Past life steal agree

13:21

with those requirements. For those

13:23

requirements when I was in recruiting worm.

13:26

I don't think directly. same. You.

13:29

Need to have above the point five. You.

13:32

Need to have a degree from

13:34

this prestigious university to succeed. I.

13:36

Think it was more. A

13:38

filter because we as a company back then

13:40

had given chances to people who than that

13:42

of people with us. We've. Given chances,

13:45

the people who were not from prestigious universities.

13:47

as it did not go very well. So.

13:50

We saw that if we change

13:52

the requirements. Towards. The upper level.

13:54

Yeah, not every candidate is going to be

13:56

good, but. More candidates are going

13:58

to be good then if they have a less

14:00

than three months I've been if they did not

14:03

go to a prestigious school. Nothing that the same

14:05

out as the I wish there were an easier

14:07

way for recruiters to filter to get the gems

14:09

to get the people who actually are good. Our

14:13

would be good at the job without having

14:15

to apply said saves. Hammer

14:17

Hammer Selter Accounting estimated

14:19

it. I

14:22

mean really, it sounds like. The companies

14:24

were managing their own risk, right?

14:27

Because to than it's risky to

14:29

accept people who may be didn't.

14:32

Do quite as well in terms of performance

14:34

in school weren't will hire the best universities

14:36

and it wasn't that you couldn't be good

14:38

at your job if you didn't have those

14:40

things, but it's just that there's a risk

14:42

to the companies seeking out there to settling

14:44

and take. Hypnosis Ecuador's those Risk

14:47

management It was. We. Have

14:49

a deadline for me up at the end of

14:51

the year. we need some. It can come in

14:53

and hit the ground running. and the risk of

14:55

hiring a bad higher? the risk of getting someone

14:57

who's not quite good enough. Puts. Us

14:59

three, four, five, six months behind.

15:01

Now we miss or deadlines and.

15:04

Here. We feel bad for the engineers who may have

15:06

been able to get there, but this is a business

15:09

and as a business we need them a above above

15:11

above the law so that's. Forty. That's how

15:13

worth of the company I was at. L

15:15

Se you did recruiting and held to do that

15:17

for. The. Do That. For about a

15:19

year. After the first six months, I saw a

15:21

job posting for the company I was working at.

15:24

That I said north. Pretty sure I could do

15:26

that. The job posting was technical support. And.

15:28

It's wasn't technical support like my laptop

15:30

is broken or my screen will turn

15:32

on. can you can help? It was

15:34

external tax for it was helping the

15:36

people who spent millions of dollars on

15:38

enterprise software that we were selling. To

15:41

help. Get. It back up and running or

15:43

to fix whatever bugs we were spiriting. So when. It's

15:46

during the morning and service on fire

15:48

under gotta build. Call to Ssh or

15:50

remote. Internet server. To.

15:52

Fix the problems and is weird because I

15:54

felt like. It's kind of what I'm

15:56

doing the whole less make as the like already. I

15:59

was a bit of a perfect and or troubleshooter. Because

16:01

I was always fix him. I'll problems whenever something came

16:03

up on a computer, whenever I'm trying to install some

16:05

game or. Upgrade Some hardware men.

16:08

I've upgraded my hardware and literally is

16:10

caught on fire. Think I've I've dealt

16:12

with a lot. This is my personal

16:14

I saw. I felt completely ready. To.

16:16

Be able to handle that export Rowan I excelled

16:19

it for fifteen years. Fifty years. Wow.

16:21

Okay so you basically did what you set

16:23

out to. d started in a non technical

16:25

role at a tech companies and you very

16:27

quickly successes is no time at all

16:29

at sea. a season long. That's it to

16:32

get a raise. you know before a

16:34

years up you Kansas and into a completely

16:36

different all that actually is taxable. Soon you

16:38

think that's your polls of having attacks

16:40

are being a programmer when you think back

16:43

to where you are as a kid

16:45

looking into your career. did that fulfill that

16:47

dream For you know. Know

16:49

not all. I was working.

16:52

To. Pay the rent. I was working to pay

16:54

my bills. I was not. Passionate.

16:56

About. That sport didn't drives

16:58

me. Well this year's is quite a

17:01

long time to do something that you're not really into

17:03

what he thinks have to units as along. Well.

17:06

I think I have a high tolerance for

17:08

doing things that owns way. I ran track

17:10

for eight years. I continued running it. When

17:12

I got the cows since it's paid for my

17:14

degree so I feel like I had a long

17:17

history already have. Force.

17:19

Myself to do something that may be isn't.

17:22

Very. Fun but this is something a you have to

17:24

do. I is a

17:26

very powerful, I don't have that skill.

17:28

I am very bad at two things.

17:30

I. Don't want it is. I think that

17:32

isn't very good muscle to us. Delta am

17:34

very happy. That said, as he turned out

17:36

that way my I think the best experienced

17:38

a concealed gave me was having that tolerance

17:40

build up. Your

17:56

somebody looking to connect with the diverse

17:58

audience of developers? Look no further. The

18:00

Deaf community is the go to

18:03

destination for developers. To learn, connect,

18:05

and support each other. You. Can

18:07

share your message with the fifteen

18:09

million developers that visit every single

18:11

month for using are powerful made

18:14

advertising. And I moved

18:16

to the. So

18:26

you did export sitting years but you are not

18:28

doing to export any more You are fully. Into

18:30

a I you Kill Found that accompany

18:33

your cheese an Officer How did you

18:35

first get interested in a? I know

18:37

Ai and tech support are both intact,

18:39

but there are wildly different industries Are

18:41

friend technical skills to friends knowledge out

18:43

as you first get exposed as. He

18:46

a good friend of mine took me for a ride

18:48

in his Tesla. And. He

18:50

activated the autopilot and took his hands

18:52

off the wheel. And a car was

18:55

is driving itself. I'm like. Why?

18:57

Is this the decision is when he

18:59

eighteen. And. I've.

19:01

Heard of self driving cars? I mean, who

19:04

hadn't at that point. But I'd never seen

19:06

it, I never experienced that. I'd never lived

19:08

it. I'd never been sitting in a car.

19:10

Seeing. Someone slow on the

19:12

highway having the carpet on the less

19:14

signal. Turn into the fast lane

19:16

past that car and then realize hey I

19:19

probably should have been a fast and anymore

19:21

Fernando might like her and move back into

19:23

the center lane. To me it was is

19:25

fascinating and I knew that I had to.

19:28

Be involved in some way. So that

19:30

is such an interesting story because I'm

19:32

sure as as A at this point

19:34

and and twenty twenty four know is

19:36

that people have been and Tesla has

19:38

tons of people have done demos has

19:40

either been in that you know car

19:42

as A was doing the self driving

19:44

or they've seen videos a self driving

19:46

cars they've read about it. It's absolutely

19:48

see this as experience but all of

19:50

people aren't. She's. A I officers

19:53

at their own start up right? They. enjoy

19:55

that experience they tell a couple friends about it

19:57

makes it a post on twitter and they move

19:59

us but for For you, it changed, it literally

20:01

changed your life. It changed the course of your

20:03

career. What do you think

20:06

it was, either about the experience or maybe

20:08

it's something about you that made

20:10

you take it so seriously and you

20:12

use it to really transform your career? Well,

20:15

yeah. After that drive, I

20:17

rushed home and I immediately

20:19

took a deep learning and machine learning class

20:22

because, again, I just felt like I

20:24

had to be a part of this. I had to understand how

20:27

the car can look at cameras and figure

20:29

out what it needs to do. It's

20:33

only just now that I'm realizing this, but when

20:35

I was a kid and I first saw that

20:37

3D video game, Wolfenstein 3D, and I remember saying,

20:39

wow, I need to be part of this. I

20:41

need to understand this. I need to be

20:44

in this world. I translated that into I

20:46

want to be a computer programmer so I

20:48

can make video games. I

20:50

think this may have been a second chance for me because

20:53

when I first saw that Tesla, when I was

20:55

first in the autopilot, when I first experienced that,

20:58

my feeling was I need to

21:00

be involved in this. I can't let this pass

21:02

me by. I need to somehow understand

21:05

how this works and be one of the people

21:07

who uses it, who creates it, who makes

21:09

it their own. I

21:11

may have been a part of that. A second chance.

21:14

That's really powerful. When

21:16

you decided that, did you have a sense

21:18

of how you were going to do that?

21:20

Were you going to go back to school,

21:22

take more courses, apply for a job? What

21:24

was in your mind at that time? At

21:27

the time, I didn't really have a defined plan. I

21:29

just knew I had to learn about it and see

21:31

where that took me. After

21:34

I took that class, unfortunately, I

21:36

didn't really get what I wanted out of it because

21:38

I didn't have

21:40

any real world data to work

21:42

with. One of the datasets

21:44

that I was most interested in

21:47

was the Titanic dataset where they give

21:49

you a CSV of a thousand

21:52

or however many people boarded the

21:54

Titanic and you get all this information about

21:56

them, how much money they make, how expensive

21:59

their ticket was. where their

22:01

family was from, all these different

22:03

details. And at the end,

22:06

you can create a machine learning model to predict

22:08

did they survive the Titanic or not? And

22:10

I could only do things

22:13

like that so many times until it's just like,

22:15

what am I doing? I could use that to

22:17

learn, but it didn't drive me. It didn't pull

22:19

me forward. It didn't make me feel

22:21

like I was actually doing something. And if I really

22:24

wanted to be a part of this new

22:26

revolution, I had to do something. I couldn't

22:28

just play with experimental 100-year-old data. So

22:31

it wasn't until Nadia and

22:33

the Storygraph came around, Nadia Odenayo, where

22:36

I finally had that chance. Tell

22:38

me that story. So

22:40

Nadia had just started the Storygraph. It was

22:43

still in alpha. Had maybe a dozen users,

22:45

nothing really publicly available yet. And

22:48

Nadia Odenayo has been my friend for many years.

22:50

I met her with you at a tech conference.

22:52

She was one of the speakers. And

22:54

I was just catching up. And I

22:57

was asking her, hey, what's the biggest problem on

22:59

the Storygraph that you have right now? What's the

23:01

biggest thing that's holding you back? And

23:03

she mentioned something that this little

23:06

light bulb in my head, I said, oh my gosh, that

23:08

class I took a year ago, I

23:11

think I can do that. And I asked

23:13

her very cautiously. I'm like, I know this is

23:15

your business, but you

23:17

think you could send me an export of your data? I

23:21

think I could solve that problem. And she was like, OK,

23:23

I don't think she expected much. But when she gave me

23:25

that export, whatever else I was doing

23:28

for the next three days, got put to

23:30

the side. I spent the next three days up

23:32

at night, barely getting any sleep, making

23:35

the best model that I could to

23:37

predict the mood and pace of a

23:39

book. So the whole thing for

23:41

Storygraph was, and still is, life is

23:44

too short to read a book you're not in the mood for.

23:46

And the only way to accomplish that goal

23:48

of matching up the reader with

23:51

the book that matches their mood is to

23:53

have a large data set of books with

23:56

tagged moods. And at the

23:58

time, she only had about 2,000. thousand books on

24:00

the story graph because she was doing that manually.

24:03

Very hard, long, arduous

24:05

manual process to do all the research on

24:07

the book, determine the moods and

24:09

then add it to the site. So

24:12

I created a mood and pace prediction model.

24:15

So now when she wants to add a

24:17

book, she just types in

24:19

the information, hits enter and

24:21

within three seconds less actually the

24:24

mood and pace predictions were printed out. And

24:26

when I first presented that monitor, my three

24:29

days of hard labor, I wasn't

24:32

too sure if it was good enough because I said,

24:34

it only has about a 60 percent success

24:37

rate right now. And now

24:39

as someone who has more experience, I realize

24:41

60 percent accuracy of something that has 16

24:44

classes or 16 different variables that it

24:46

could be actually pretty good. But back then

24:48

I was like, it's only good about 60 percent.

24:50

And I showed her the output of my model

24:53

and Nadia looked at it for a few minutes. And she

24:55

said, Rob, this model is better than me.

24:59

I said, Rob, well, I'm looking at some of

25:01

these. And like this book right here, it literally

25:03

has mystery in the name, but

25:05

I didn't tag it as a mystery book. I

25:08

don't know why, but your model did. Your model

25:10

picked up on that. So for

25:12

the next few months, I worked with Nadia on

25:14

the story graph to add as

25:16

many machine learning, as many AI features as I could.

25:18

We started with the mood and pace prediction model. Then

25:21

I made a book recommendation model. And

25:23

now a few years later, I have about

25:25

15 to 20 machine

25:28

learning models. That are live in production, running

25:30

on servers that millions of people are using.

25:33

And it all came from that one

25:35

Tesla ride that I took that sparked inspiration

25:37

from maybe my failed dream as a kid to

25:40

be a video game

25:42

person and having a second chance

25:44

to now taking the class, seeing an

25:46

opportunity and jumping on it. Wow,

25:48

that is incredibly inspiring. So it really just came

25:50

down to that one class that you took. So

25:52

is it really that simple? Can I just take

25:54

that class that you took? Or was it a

25:56

cost, by the way? It

25:58

was the fast AI. Jimmy Howard.

26:00

He's a fantastic teacher. Do it. He teaches

26:03

accede. I think it's pretty cool that he

26:05

explains it as. If. You're teaching someone

26:07

to play soccer. You. Don't start by

26:09

giving them a pen and paper. And

26:11

explaining the physics of ball mechanics and

26:13

aerodynamics. has it's going into the net

26:16

Know you? just you get a soccer

26:18

ball, Gloucester, fielding does kicking around and

26:20

u cel by doing. That's. How

26:22

he teaches Ai he teases ai by. Within.

26:25

Your first class you're going to. Great model and it's

26:27

gonna be a simple model. May not be the best

26:29

but you know what? Three years earlier

26:31

it would have been a state of the our

26:33

best model that even Google could have created because

26:35

the tech is advancing so fast. So in some

26:37

ways yes, it really is that simple if you

26:39

just want to learn. The. Task:

26:42

It's want to learn the skill set. It

26:44

can just be that simple. You may want

26:46

to know little bit a Python before you.

26:48

Start. The fast Ai course specifically, but there

26:50

are other courses you can take. That

26:52

maybe not require as much python or you could just.

26:55

Spend. A few weeks or months of learning

26:57

Python and then you'll probably excel in the

26:59

fast Ai courses Wealth and Fassa course is

27:01

actually. A pretty stable industry. Since

27:04

taking the course, I've read that every Tesla

27:06

engineer. Whose. Work on autopilot is

27:08

applied to take the course. Think every Google engineer

27:10

who's working on a I is required to course.

27:12

It's a free course. So it's kind of funny

27:14

that the. Course. That maybe

27:16

some of those test engineers took

27:19

to create autopilot. Or. To help

27:21

them understand how it works. Is.

27:23

Also, what got me into a I. Saw.

27:25

In terms of learning the skillset Yeah, it's

27:27

as simple as taking a course and I'll

27:30

think of requires. The. Deathly doesn't require

27:32

a Phd. It is all fire Advanced

27:34

Calculus. It doesn't require years and years

27:36

of research in order to use the

27:38

tools Now to invent the tools. Yes,

27:40

if you're out there and you're tied

27:43

to a descent, the next machine learning

27:45

paradigm, the next large niggers model advancement.

27:48

If it's on to compete with set Cbt at

27:50

creating chat cbt from scratch you. Probably.

27:52

Do obesity hobby? do want to go did x to

27:54

school and but. That's the same

27:57

way as if you want to invent a new

27:59

javascript, you may. It appeared to be something you

28:01

want to vent the new wooden rails. That's That's not

28:03

what engineers do. We use the tools that are available

28:05

to us. and right now. We've. Reached a

28:07

level where the tools are easy and accessible

28:09

enough. And a hardware is good

28:11

enough that almost anybody would a laptop.

28:14

And. A few weeks of time.

28:17

Can learn how to make. State. Of

28:19

the our models, I think that's pretty cool. That

28:21

is really cool and it sounds like to me.

28:24

There's almost two camps as

28:26

has two sides. As a

28:28

I there's the ai that

28:30

has been created. The people

28:32

are open a I who

28:34

are building taxi to see

28:36

who are building dolly. And

28:38

they're creating these tools. For

28:40

the rest of us to use. And

28:42

it sounds like you're saying if you

28:44

want to be on that side of

28:46

the table then yeah it's having to

28:49

go to Mit needed or savory together

28:51

Pct to do that, Calculus three or

28:53

wherever that as you probably need to

28:55

buckle and and thus and time and

28:57

to an education and into getting a

28:59

procedures degree. But if you are just

29:01

trying to build with a I, if

29:03

you want to incorporate it into your

29:05

product, you wanna leverage it to create

29:07

something big and beautiful. You don't need

29:09

all those athletes. And all those things. So

29:11

I guess my question for you is. We've.

29:14

Heard a lot, especially this past. Year

29:16

And it's Wine Twain. Three, we've. Heard a

29:18

lot of people making fun of apps

29:20

that are just as Cbt rappers right

29:23

people who just plug them the A

29:25

P I N yards of have a

29:27

little tap out of their own and

29:29

and basically or to sending data to

29:31

techy Be here when you talk about

29:33

building A I tools and products in

29:36

the things that you do is that

29:38

what we're really talking about or was

29:40

had unless. I'm indifferent. Oh, definitely not.

29:42

That's the easy way out. and if

29:44

all you're doing is using someone else

29:46

is a P I. For. A

29:48

core feature of your app. The. New

29:51

York do anything special or unique in

29:53

that your competitors can top you very

29:55

easily, so you didn't do something a

29:57

little bit beyond. To sending

30:00

data to open a I or any

30:02

similar company will we specifically do serve

30:04

as the vast majority of our models

30:06

are not large think was models. But.

30:08

We do have a large they was

30:10

model feature set that specifically is called

30:12

the Story that preview. When. You

30:14

visit a book page you'll see a sort

30:16

preview of what type of read or that

30:19

book would be good for and that is

30:21

powered by a local self hosted model that

30:23

I personally am hosting the i'm not using

30:25

an external eighty ice for that and frankly

30:28

with the popularity of the story graph. When.

30:30

I did the math if I were to use

30:32

a Pr for that, which I never would because

30:34

of. Data. Privacy issues like our

30:36

users are not be happy with me sending

30:39

their dad off to it and externally the

30:41

I like that but if I did want

30:43

to do that, food costs seventy five thousand

30:45

dollars a month a even with aggressive casting

30:47

and. I'm hosting that model

30:50

for a fraction of the cost and

30:52

says I'm hosting the model I can.

30:54

Do. Some special things: I can fine tune

30:56

it. I can modify it specifically to the

30:59

dataset and users that I need. An.

31:01

That. Model is mine and nobody can take that for

31:04

me. And it's not easily copy evil and even

31:06

if someone tried to copy it, they wouldn't have

31:08

access to the things that I have access to.

31:10

To copy it so I feel like

31:12

the next paradigm and terms of. Building.

31:15

Your own A as features or the

31:17

next paradigm in terms of integrating ai

31:19

into your app is not using an

31:21

external A P I, but it's something

31:23

more custom. It's something that yourself hosting

31:25

a something that you have control over

31:28

and your users. Your customers can. Feel.

31:30

Confident. That. Their data is

31:32

not been compromised. Okay,

31:35

So we're not building our

31:37

own lives language models were

31:39

using something that someone elses

31:41

bells. Where does the Ai

31:43

officers skill set com what

31:45

about what you're building how

31:48

you're using it is the

31:50

part where your leveraging your

31:52

knowledge and your talents as

31:54

an engineer. Sit at there.

31:56

A few things that I think about. when when I

31:58

hear the question I think about. The.

32:00

Six Months. Literally Six months that it took

32:03

me working day and night. Focused.

32:05

Only on the store that preview. Were

32:07

yes I do not creep. With.

32:09

Millions of dollars, a large eggs model from

32:11

scratch. But. I still needed the

32:13

up to be good. I still needed to

32:16

create something that are users would. Derive

32:18

enjoyment from and that took a lot

32:20

of sized at so I would say.

32:23

In order to create a good as

32:25

feature in addition to the knowledge of

32:27

this understanding in general how it works

32:29

your car's patience, some was like that

32:32

Thomas Edison I know, ten thousand ways

32:34

that won't work to create a light

32:36

bulb. I experimented with thousands of different

32:38

ways of creating that feature. And

32:41

I don't think that even. The.

32:43

Beach these at open A I who create

32:45

amazing lodging was models l think they could

32:47

have stepped in and done that any faster

32:49

than I did because I have the knowledge

32:51

of what my customers what I have the

32:53

knowledge of what it means to have a

32:56

good book Recommendation: And they will

32:58

have their contacts business Now they do. So.

33:00

I think that. It

33:02

can be overwhelming or it can be

33:04

a little bit intimidating to think that

33:06

of competing against these giants in the

33:08

world to are creating amazing technologies. but.

33:11

Their. Knowledge is limited. In. Ways

33:13

that. Yours. Isn't you can

33:15

still provide value? By bringing

33:18

your industry expertise in whatever it is

33:20

you're working on. And do in

33:22

a way that no one else can if you're willing to spend

33:24

the time. Telling

33:26

more about the ten thousand ways

33:28

because it it almost feels like

33:31

building. With a i'm trying to

33:33

create machine learning features the are.

33:35

Fundamentally different from the on the web

33:37

developer and I think about building a

33:39

web app. A lot another way

33:41

that do it. there's a pretty be no

33:43

clear is that ones that is that Three

33:45

says area have always had do that but

33:47

as as if you're using framework like rails

33:49

what has a lot of opinions There's definitely

33:51

best practices that you just kind of following

33:53

You remake said enough to make something different

33:55

you put together a different order bucks here.

33:57

So car generally. Following a person. After

34:00

that already exists, but it seems like gear.

34:02

Ten thousand ways to sail feels like a

34:04

completely different paradigm. Tell me what it's like

34:07

to build an Ai feature. The.

34:09

Process to build be feature that has the

34:11

laws that was model the start of preview

34:13

feature. Was. Not that different

34:15

than a process of building beads. Very

34:18

first model I made a story that

34:20

the Mood in Peace production model where.

34:23

There's is mean that I love where it's somebody

34:25

with a long stick in a big pile of

34:27

garbage at the bottom. Added a stirring the garbage

34:30

seven stopping and second what the output is and

34:32

see if it's something that's good enough. it's not

34:34

good as keeps During. And the

34:36

title was life as an A Engineer.

34:39

That's. Pretty much what it feels like you are

34:41

experimenting, defined what worse and with Mars Bengals

34:43

models. Yeah, there are a lot

34:45

smarter than the models that we've seen in the

34:47

past, but. Is you don't know

34:49

exactly what to give it and exactly how

34:51

to structure your pants, what settings to use them

34:54

in? Even if I just thought about the

34:56

settings, there are thousands of different permutations of the

34:58

settings for the last thing as model the

35:00

you can have and then you have to think

35:02

about the inputs. Yeah, think about how long

35:04

as the output, what type of output be a

35:06

game is? Forget about cynical stuff completely. What

35:09

type of output would a user expect

35:11

to see from visiting this book? Date.

35:14

Is this to type of output that is

35:16

even possible to Las Vegas model is that

35:18

I'll put valuable because they can still read

35:20

the book description right there. Why?

35:22

Would you give them another book this gibson? yeah

35:25

to do something different in order to give them

35:27

value. A second part of that feature is called

35:29

the Story Grass personalized previous where it looks at

35:31

all of the books that you've read over your

35:33

entire history on Story Grass. And it

35:35

uses. That information to determine what type of books

35:38

you typically like to read and then it looks

35:40

at the book, pays they are on and says.

35:42

Would. You like this book? Why or why

35:44

not? That's the featured at the longest time

35:46

for me to work on because. People.

35:48

Have so many different book tastes. People are so

35:51

many different things that they look for and. That's

35:54

something that. I was only able to

35:56

do. Because. Has already been doing

35:58

this for years. Have already been working on

36:00

candid book recommendation models for years. Already have

36:02

a good sense of. What?

36:05

People want in a book app. Out

36:07

even see that's more important than. Having.

36:09

The knowledge of working on models for years.

36:11

this knowing. What? People want and

36:13

what they spoke to see what they would

36:16

arise. Are you from. Saudi.

36:18

Like eighty percent of it. Oh wow,

36:20

that's very interesting. So. Some of what

36:22

you said reminds me as Pawns Engineering

36:25

that's a concept of it's as a

36:27

whole feals. That's something that I remember

36:29

being really really popular. Sicilian taxi be

36:31

t under a P I became available

36:34

and golfers rebuilding south. They were courses

36:36

and staff and lots of conversations. Round

36:38

Engineering. How to ask is

36:40

in this context. Teddy Bt

36:42

the right question with the

36:44

right information. And giving it the right

36:47

combination to get the answers that he wants?

36:49

Is that what we're talking about? Or is

36:51

that part. Of the story. That's

36:53

the funny part of it. but there can

36:55

be the whole story. and when you think

36:57

of the story of has millions of users

36:59

and honest or that there are millions of

37:01

books and every user is unique. Every

37:04

book is unique. So. I

37:06

know how many trillions of combinations of that

37:08

can there be? There is a bit of

37:10

prompt engineering there, but. You need to

37:12

figure out a way. That. Even.

37:15

with the same prompt. It. Would still

37:17

work for the majority of users with the I guess

37:19

there could still be considered Prada engineering but it does

37:21

feel that with me because it's not. Only. To

37:23

prompt that I'm dealing with dealing with

37:25

so many other variables. Personally,

37:28

Elements: mostly the setting so there

37:30

are only get to turn a

37:32

corner, spin doctors, the temperature setting

37:34

with toes the model essentially how

37:36

random shit Id. And. If you

37:38

set the temperature setting really high, It

37:41

sounds. More like a regular person

37:43

talking but can say some crazy

37:45

stuff has it. He said it

37:47

really low. then. It almost sounds

37:49

like a robot, but you can feel confident that

37:52

it's not going to go off the rails and

37:54

start talking about out a no Flowers when it's

37:56

the book has nothing to of hours so. When

37:58

I save a car space and. Ten thousand

38:00

ways. It's about combining. All.

38:03

Those variables, the prompt abiding the information

38:05

you have on what the user space

38:07

or with the user specs, the seats

38:10

on a different variables and haven't even

38:12

gone into. The

38:14

server that can be hosting this day and how

38:16

fast it as server need to be were software

38:18

going to use the hosted where is the server

38:20

gonna be and six six was. Coming

38:26

up next Map shares my his theory work

38:28

with a I would. Say

38:31

that it is Cindy started after

38:33

that. The

38:44

thing that I find very intimidating an

38:46

overwhelming about Ai as someone who is

38:49

not an aunt who doesn't actually do

38:51

it as part of their jobs is

38:53

just the constant updates as feels like

38:55

every day there's a new model and

38:57

is setting a new company and me

38:59

start up a new Thunder seems to

39:02

be just things changing so rapidly that

39:04

I have no idea how I would

39:06

even keep. Up with the industry in a can

39:08

be very overwhelming to think oh my. Goodness, I got

39:10

him what happened you know, a year ago and

39:12

I learned that the new everything will. That's how

39:14

do you do that? How do you keep up

39:16

their knowledge? given the constant. Changes Information has

39:19

so much one. This

39:21

is my hobby is when I have a

39:24

spare minute I'm searching through. I'm fine how

39:26

the latest news I can find on what

39:28

latest models are was coming outward. Bullied techniques.

39:31

How can I implement this and start

39:33

out like this? This is how allowed

39:35

the features on Syria started that way

39:37

where it wasn't even in my head

39:39

until be randomly searching. Found an interesting

39:41

article in said whoa. I

39:43

can do that. I can add that assert f

39:45

I think our users would love that and. That

39:48

might not sound like an inspiring answer

39:50

were oh great Academy does my hobby.

39:52

I don't think so if you find

39:54

something that works. And. You stay

39:57

on. That version is not like yourself

39:59

Ruger any worse. If you don't to the

40:01

upgrades for the latest version if he if a

40:03

new model comes out and he don't upgrade to

40:05

at your existing models to bed is still going

40:07

do a nice to do but i find that.

40:09

An industry that is evolving so quickly. As

40:11

he said, Having like

40:13

your finger on the pulse of it. Having

40:16

a good sense of what's coming, having a good

40:18

sense of what's already out, having a good sense

40:20

of the latest techniques is very vital. I mean,

40:23

I wouldn't look up the latest news

40:25

for with and rails are javascript or

40:27

anything like that because comes capacities again

40:30

and side javascript and will be. Careful.

40:33

I know I know they are doing amazing work

40:35

and you doing a lot of great things but.

40:38

It doesn't feel the same where. I.

40:41

Think if I go week or two without seeing big

40:43

news and a i'm Michael what's going on with I

40:45

did. I miss it and must be out there because

40:47

things are happening so fast. It's not like that with

40:49

other frameworks. It's not like that was asked to read

40:51

on rails or anything else. So I do feel like

40:53

at this point if you do want to get the

40:55

most out of it. At least have. A

40:58

few blogs are a few. People. On

41:00

social media that you follow this to

41:02

keep up with the latest news because

41:05

he might be instantly couple. Newsletters,

41:07

Any that a Reddit? a lot to

41:09

sign? The latest technology isn't that? Latest

41:11

updates as well. So is there folks

41:13

listening who are also inspired by your

41:15

journey about the south? Are you new

41:17

Zero Ai you know? Took a ride

41:19

and then took a course and then

41:21

basically have this app as freezing Learns

41:23

that people a month. People who inspired

41:26

by that? How do you recommend They

41:28

get started. But an. Easy way to get

41:30

started is to. Take. The same course

41:32

I took the first day. I course. I think

41:34

that's. A fantastic way

41:36

to very quickly. Get. The

41:38

ground Running within the first course with in

41:40

first class you'll have a model that you

41:42

can play around with an. Experiment

41:45

with and see where it goes. I think the main

41:47

thing I can say when you going to destroy any

41:49

is. To have an open mind.

41:51

and if you already experienced engineer, you're probably

41:53

used to things not working the first time

41:55

you try. You're. Probably used to running

41:57

your app. Seeing an error. Called back. Then

42:00

in two hours to fix it is found the sixth and now

42:02

you move on. With. A i

42:04

the penny away his I do you might be stuck on

42:06

a for a lot longer you might have a lot more

42:08

a to reasons and depending on what you're trying to do.

42:12

You. Might not easily find the answer Online chats

42:14

you bit. he may not give you the immediate

42:16

answer Google or may not. Know exactly

42:18

how to help you because things are changing so

42:20

fast. So I'll just say. To.

42:23

Expect things to break. Keep an open mind.

42:25

But. Just. Keep. Pushing because.

42:28

So. Often than I do think. I.

42:31

Get enjoyment out of a I in a way that

42:33

dinner can into amount of programming. In the past. Owns

42:37

was driving me. As

42:39

you orleans conversation as you were living

42:42

the dream when you first technical job

42:44

nor does in the industry you see

42:46

doing the work itself. As a tech

42:49

support engineer he. Said. He

42:51

said that it wasn't It wasn't a dream

42:53

that all. What about mouth disease? You living

42:55

the dream say? Oh I'm living the

42:57

dream. I literally false the both my laptop in

42:59

my bed I'd wake up a three the morning

43:02

excited to see for my latest experiment is going

43:04

every day. When I have a spare moment I'm

43:06

checking the latest news checking help people liking the

43:08

fetus on story grassy how I can improve them

43:11

the everyday is a new talents every day's a

43:13

almost almost like a new world on a weekly

43:15

basis with the new news that's coming out so

43:17

yeah I can I'm as having any better than

43:20

us that's like five years as Isis on this.

43:30

And that the another episode we ask our guess

43:33

is on the planes as enzyme. For instance,

43:35

Rob I read this on the list

43:37

I n. Number. One at

43:39

worst suffice I've ever received as. Worth

43:42

of I serve ever receive is there

43:44

has to be. If you're

43:46

not good at calculus or math in general, that

43:48

you can't be a good quarter and. I'm

43:51

living proof That the sad truth is I know many

43:53

other engineers who are also living proof. I think if

43:55

you ask most engineers how much cause they do on

43:58

a daily basis. That

44:00

a bang smoking at all. So I

44:02

think that. My. Life with on

44:04

a very different path. Because of that. that advice

44:06

and I'm happy with where ended up. But.

44:09

Oh people. Now based on condenser. Number

44:13

Two: Best advice I've ever received is.

44:16

The best advice of ever see this did to save

44:18

money. If it wasn't for the fact

44:20

that me and my partner. You.

44:23

See. It was a bit

44:25

of fact that me, my partner we're. Saving.

44:27

Money: Were explicitly deciding to live

44:29

below our means. To be

44:31

when to be a thing or think gonna be

44:33

never would have been started. Story graph: I never

44:35

would have been able to vehicles on the for

44:38

store. Guess it's the fact that. We.

44:40

Had that as a buffer than you don't

44:42

need that much. Sometimes you, and sometimes you

44:44

may only need a few months of money

44:46

saved up so that you can take the

44:48

risks that you need take. But I went

44:50

years on storing f not making a salary

44:53

because we would have an income and. I.

44:56

Knew my job for is because I

44:58

knew that. In this house

45:00

dedicated myself to at one hundred percent. I

45:03

would not be able to. Create.

45:05

Technologies are not be able to quit, The models

45:07

are not be able to move it to where

45:09

it needed to be. To. Be a successful

45:12

company. Is no active the now

45:14

the side it is required to it's time required to

45:16

it's thinking. Then. I

45:19

didn't have the money saved up. Do that. Or.

45:21

More specifically, if we couldn't live on one of

45:23

our salaries like you don't need to have a

45:25

year of money saved up if. You. Can

45:28

live off one salary. So one thing that we

45:30

did when we first got together as we decided

45:32

to only live on one person salaries of the

45:34

other person loses their job. You.

45:36

Can still survived it still pay the

45:38

rent. the network great for you when

45:40

you took. Three. To six months off

45:42

to do. According to camp, he didn't have

45:45

a Saudi during a time opposite you're paying

45:47

money for has given. I might have Tesla

45:49

camps and. That. Put you on a

45:51

path. And when it was my turn to put me on

45:53

a path, To My best advice that

45:55

I get people nowadays is to. Try

45:58

to live below your means so. When

46:00

opportunity comes. You. Can

46:02

actually go forth and.

46:05

Never. Three, My first coding priceless about.

46:08

My. First coding project was. In.

46:11

College I think I was a sophomore in

46:13

a was Tic Tac Toe game so I

46:15

had to figure out how to draw a

46:17

Tic Tac Toe map added accept input from

46:19

a user to put the x's and o's

46:21

in the right places and how to write

46:24

a computer program that can beat the user

46:26

at Tic Tac Toe which. Was.

46:28

Actually, less a lot of fun. I look back on

46:30

those times. family. Member

46:32

For one thing I wish I knew when I

46:34

first started to code is. I

46:37

wish I knew. To. Be prepared

46:39

for errors. One of my first

46:41

college. Projects. That I worked

46:43

on and pulling an all nighter for

46:45

those five in the morning. And I

46:47

had one hundred and thirty three. Eris

46:50

and would you believe? When.

46:52

I found this cause it was because

46:54

of one semi colon those missing these

46:56

sets it. I added that one semi

46:59

colon and my husband a dirty air

47:01

is went to zero and. Now

47:03

I know that it's. It's weird things

47:06

worth the first time. That

47:08

might actually be bigger problem now on

47:10

the senate or the first time he

47:12

tried an embassy think of very tassie

47:14

him as very sentence and tested you

47:16

tube or his science youtube channel. Where.

47:19

He makes videos about. These

47:21

series challenging math problems that

47:23

has been unsolved for hundreds

47:25

of years in some cases,

47:27

and how. People.

47:30

Live their entire lives trying to solve

47:32

this problem and failing. And that's not

47:34

where we are in tech in coding.

47:37

With. All the problems are solvable. He does

47:39

haven't gotten there yet. And

47:41

I think that for me personally, it's helped

47:43

that I had a background. Almost

47:45

as a professional. Athletes

47:48

to that's. Solicitors,

48:04

One of the demented with the community

48:06

find us on Dez at Gas.t L

48:08

A/code newbie team were always posting new

48:11

discussions and content related to coding careers.

48:13

And if you like the so make

48:15

sure to follow us a meet a

48:17

review on your preferred platform so we

48:19

can see making the. Things.