10 AI Skills You Didn’t Know You Already Had, With AI Architect Ramona Truta!

Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Chris (00:03):
Welcome back to Data With Direction.
I am here again with Ramona.
I'm super excited to have herback.
Today we are gonna be talkingabout the 10 AI tools that you
need, uh, skills that you needas a data engineer and how that
is gonna future proof yourcareer and probably how you
already have some transferableskills that are gonna help you

(00:25):
along the way.
So Ramona, if you'll telleverybody a little bit about
yourself, uh, just for those ofyou that may have missed the
last, uh, the last episode, anduh, then we'll jump into it.

Ramona (00:37):
Hi Chris.
Hello everyone.
Um, I'm so excited to be here.
Well, last time, uh, what we hada conversation with Chris.
We absolutely didn't finisheverything we wanted to chat
about.
So, uh, when this opportunityarise to join him on the show
again, of course I'm here.

(00:57):
Uh, I love, uh.
Talking about all things dataand ai.
So this is an amazingopportunity to do so.
I have built my long career indata and trans, and of course
there is no AI without data.
So that was the seamlesstransition to me.

(01:18):
And, uh, today we're gonna sharefrom our experience and about
the transferable skills that adata engineer has to move into,
uh, the field of ai or well,assuming everybody's already
digging their toes and some maybe, uh, are already swimming in

(01:40):
the AI waters.
So.
Take it from, uh, let, let's,let's go and have some fun and
share from our, uh, perspectivesand experiences.

Chris (01:51):
Yeah.
I think some people get throwninto the deep end with ai, but,
uh, yeah.
Let's, uh, I do wanna share thisLinkedIn post that you posted
this morning, uh, because Ithink it really applies.
It, it's, uh, you know, talkingabout enter the AI janitor, the
most needed role at the end ofthe full life cycle vibe coding.
Um, so tell us a little bitabout, you know, this post and

(02:15):
kind of, you know, it lookslike, you know, VIN post, this
great Superman post talkingabout vibe coding and you know,
what that kinda looks like.
And, uh, tell us a little bitabout what inspired this post.

Ramona (02:26):
Okay.
So it's, it's a bit of a story.
If I, if I, uh, I, I'm gonnamake a little bit of detour and
get to get to the story.
Um, Kasi ov, um.
Coin the term of data janitor afew years back.
And I wrote on this, uh, topicmyself, like the golden toilet

(02:49):
metaphor.
That is somewhere on my, uh, on,featured on my profile.
So it's this idea that, andanyone here can attest to that,
right?
That as data people and I, Imean the entire landscape, not
just engineers, right?
We all had to deal with a lot ofdirtiness in data.

(03:11):
And most of the times, problemsthat, uh, appear in, in the
data.
Uh, in the data work are reallyrooted into the quality of data.
So I, there, there were a lot ofdumpster fires that we had to
put out through our careers, andI think we, we, uh, talked

(03:31):
actually a lot during last timeright.
About data quality and someworkflows and, uh, patterns to
follow.
So, I'm, I'm not gonna repeatthose, but so Cassie's point was
that, uh, you know, she is, uh,she created an entire field, the
decision intelligence field.

(03:52):
So she has been, uh, at Googleand she, she's a prominent face
in, in, in the AI space, and shenoticed that at every, and, and
she's, um, she's a world class,uh, speaker, right?
So she's invited to the biggeststages, uh, in the world where,

(04:13):
where she shares her experience.
And she, she talks so.
Um, she noticed this patternthat when she would talk to
various leaders and variousteams from across industries,
and she asked this question, whoshould be in charge?

(04:33):
Would the data quality, the dataclinic, all, all those, you
know, the digging into that sideof, of the, the, the core of the
matter.
And, uh, she would get variousresponses.
Basically, nobody really wantedto, uh, take agency over doing
that part of the work.

(04:54):
And then she, nobody, no.
So everybody wants good data.
A good quality data, but nobodywants to have to deal with it,
right?
Because that's not a really,that's not sexy kind of work.
So it, it, it's not flashy work.
So she coined this term for datajanitor and thought, okay, if

(05:14):
we're thinking about creating,um, a specialty in, uh, uh, in
university, right?
Or in a college, let's say adesignation, uh, and she said
she, I'm gonna quote her, Hey,uh, mom and dad, I've decided on
my, um, degree, I'm gonna be adata janitor.

(05:35):
So how does that sound?
You know, imagine the parentsdropping faces, so that's why
we're dropping so much money onyour education to be a data
center.
So it it's this idea that.
The, the work requiresjanitorial, the cl, right?
So if we think about that, thatthat's really what the, the job
in entails.

(05:57):
And, but it sounds awful whenyou put it like that, but that
is such an important role.
And of course nobody will coinit that way.
So I took this to, to the AIfield, and, uh, I really, so I
started talking about, uh, thevibe coding, uh, side effects of

(06:17):
things, right?
Because everybody now is anexpert and vibe, coding and
whatnot, but it's, it's really adangerous, um, activity if you
don't know.
Much about the software, thegood software lifecycle,
software development, life, uh,lifecycle.
And I know when I white coded,uh, the apps that I worked on,

(06:41):
I, I, I stayed in the loop andeach time it generated
something, I really went anddouble check.
And you, you know, I knew whatwas happening.
So I didn't just acceptedeverything thrown at me.
And this thought formed thateverybody, and, and it's not
just my thought, right?
Many people talk about the AIslope and

Chris (07:05):
yeah,

Ramona (07:05):
somebody eventually, and I think jewelry compares it to
the big Pacific, uh, uh, garbagepatch.
It's this huge, huge area in thePacific where all the plastic,
right.
Uh, accumulated.
So, um, so thinking of, youknow, we've seen this movie

(07:28):
before and this time it's just,I think it's way worse because
everybody has access andeverybody can just do it.
Your, everybody can speakEnglish, type something, and
then boom, you have an app, andthen you generate a slope, but
you don't know what you'veinstalled, you don't know the
packages, the damage, uh, thatthat will cause and Right, that

(07:49):
will s seep through the systems.
And there's a lot of researchcoming out mentioning, uh, these
things and even, uh, ex expertdevelopers who install packages.
And, uh, it's such a tinydifference in the name right,
that you, you don't know whenyou're installing the bad one.

(08:10):
So.
Uh, long story short, uh, whenVin wrote about, I mean, there,
there were several posts righton this idea of white coding and
what happens and on the ai, onthe AI sub generated, and it's
been constantly on the back ofmy mind that somebody will have
to clean this.
Somebody will have to dug theirheels and actually make sense

(08:34):
or, you know, but how exactly doyou do that?
And what will be that role?
And I've said it a few monthsback that I hope cleaning this
slope will not become, uh, areal job, but it turns out that
it will.
And yesterday, I, I throw theterm around the ai, um, janitor,

(08:56):
and today when VIN wrote that,it, it was just perfect because
it's the vibe coding lifecycle,right?
And then at the end of it,there's so much slope and so
much terrible stuff.
Somebody has to come and cleanit.
So enter.
AI janitor.
And, uh, I immediately thoughtthat people will think of it as

(09:17):
a joke.
So then I threw in a little bitof strategic, uh, questions,
right?
That it, to, to make sure thatit's, it, this is serious.
And, uh, companies throw so muchbudget, so much budgeted AI
projects, right, without some,uh, clear objectives.

(09:37):
And if you, if they allow theiremployees to just wipe code
everything because, uh, this isthe hype and this is the hot
stake right now, they shouldreally consider putting money
aside for an AI janitor role.
And I'm so not kidding withthis.
And as a solution to theproblem, I offered to train.

(10:00):
Uh, the professionals, right?
And offer them the possibilityto upskill nothing into just
prompting an ai, but actuallyupskill and under and get real
data literacy, AI literacy.
So then you understand whatyou're doing and the system that
you are, that you areinteracting with a lot more.

(10:22):
So then you avoid a lot of thepitfalls that an AI janitor
would need to clean after you.
Um, so if this was a long wave,uh, you know, winded answer, but
it's not, it, it's a verycomplex thing, right?
So I had to pro, I felt like Ineeded to provide some context
to understand, because I use ai.

(10:45):
It's, it's a fantastic tool andit's augmenting my work and my
life.
So this is a good thing to have.
We just have to be responsiblewith it.

Chris (10:55):
Yeah.
And, and I think, um.
You know, kind of, you know,that that comparison between
data janitor and AI janitor,like, it's, it's all, um, you
know, we've got people, youknow, humans that are generating
stuff via ai, you know, via thisAI assistance.
And it's the same concept,right?

(11:17):
It, it's either way stuff has tobe cleaned up.
There's, and this is neverending, right?
Because things and people areimperfect and data janitors will
always be needed, right?
And things will always have tobe cleaned, cleaned up.
AI janitors.
I, I think that, um, you know,vibe coding in particular, ai,

(11:40):
uh.
You know, is going to continueto increase the work.
Like we might see thistemporary, you know, decrease in
the amount of work because ithelps us get from like zero to
one.
But then we also have peoplethat are out there, vodka,
coding, um, some, you know,people getting it in, you know,

(12:00):
accidentally falling into rolesand like they're completely
using AI to do their job withoutan in depth understanding of
what's going on.
Um, and, and there's gonna be alot of cleanup that has to go
on.
I mean, it's, it's, it's growingpains.
And, and, and I think that likeeverything else that's new at

(12:22):
first, it's gonna be this, it'sgonna feel like this
insurmountable amount of workthat properly will continue to
come up.
And then over time it will kindof normalize, right.
Um, but yeah, that's my, I Ithink that this janitor, uh, you
know, analogy is perfect forthis.
That's, this is,

Ramona (12:39):
and you heard it first here on this podcast, folks, so
now you can go and advertise toyour companies.
And I, I, I'm actually, I, Imade it sound like a joke, but
this is, this is a seriousthing.
And, and, uh, thank you Chrisfor, uh, rolling it back to, to
the DA data janitor.
Uh, and we, we, I think weaddressed several of the

(13:04):
pitfalls last time we, wechatted, right?
Um, because folks don't knowwhat they're doing, and
especially with ai, uh, they.
It's, uh, it'snon-deterministic.
So you, you really don't knowwhat you're doing and you're
entrusted with a, a lot ofpower.

(13:24):
Uh, and then it, it takes time,right?
Because you don't observe theproblems right away.
And, and this is a point I madeyesterday in a, in a separate,
uh, post that you, you reallydon't see the effects of what,
of the wrongdoing.
It, it takes some time, and ifyou don't monitor it, you won't

(13:45):
know that it's happening.
So then, uh, how do you traceback the origin of the source,
right?
And whose head is gonna fallwhen that happens?
So,

Chris (13:56):
and, and that's, I think that's gonna be the thing too,
is that as data engineers, uh,and as, as AI and data, you
know, architects, we understandthe need for things like
metadata.
So that things can be trackedback.

Ramona (14:12):
Yes.

Chris (14:13):
And people that are out there vibe coding, there's
nothing in that vibe code.
I promise you.
There's nothing in that vibecode unless you ask for it.
That is tracking any metadata.
It's, it's just not happening.
Right.
Because you're not asking for itand the AI is not gonna, you
know, self prompt and be like,Hey, you need, probably need all
these other things here.

(14:33):
Um, it's there.
Yeah.
So, uh, so I, I think that'sgonna be one of our biggest
challenges as we clean up someof that, as we become those AI
janitors.
And because it will be, I thinkit'll be a lot of the data
engineers and, uh, dataarchitects and, you know, those
of us that have historicallybeen in these roles that will

(14:54):
have to come through and clean alot of this stuff up.
Um, it, the difficult piece isgonna be.
Tracking it back'cause therewon't be any stored metadata
that has been, has been there.
And, and in some cases nohistory, right?
There's no, you know, historicaltracking of what's been going
on.
And so important for those ofyou that are out there, vibe

(15:16):
coding, so important to, youknow, for you to integrate these
concepts into what you're doingso that you can, I like this
term, inspect what you expect,right?
Mm-hmm.
You, you want to make sure thatyou have the capacity to inspect
what you expect from the thingsthat you're creating.
And to do that you're gonna needto do things like metadata

(15:38):
tracking.
Um, but yeah, but

Ramona (15:41):
to, to even get to that point, you need to know what you
don't know.
So you need to, to have theawareness that there are a lot
of things that are, that shouldbe in place that you don't even
think about it, or you, you justdon't know, right?
Because you, you, you.
No, nobody, uh, you didn't havethe opportunity to learn or to

(16:03):
be aware that you need thosethings.
So it's, it, it's a lot.
And, um, yeah,

Chris (16:10):
so we have a couple of comments here that I think
really apply, uh, IL thatwelcome, uh, nickel's, one of my
longstanding, uh, podcast, uh,viewers.
Hello.
Been here since the beginning,pretty much, uh, coding with the
vibe of someone who has anoslevel power, uh, but not the
brains to match, right?
I mean, you just, just like, youknow, like we're talking about,

(16:32):
you don't have the, you don'tknow what you don't know.
You don't have the history toknow

Ramona (16:36):
tunnels after he, he got all the re all the stones right
on.
Uh, yeah.
Not before, after he, he, uh,he, he got all of them.

Chris (16:47):
Uh, and then Robert, uh, who's awesome, uh, Ramona makes
great point on vibe codingwithout proper dev experience.
Um, and you know, he's been.
Making sure I'm not using,losing the same words.
Uh, yeah.
Proper dev experience has beengenerating software with
security.
Yeah.
I mean, security's a hugeconcern too there, right?

(17:07):
Um, I, I've even seen vibecoding come out and it's put
the, you know, things like APIkeys in the, within the script,
and, and that's something youdon't wanna do, right?
Usually, you know, depending onwhich of the tools that you're
using, uh, some are better thanothers, you know, uh, cloud code

(17:27):
is a little bit better aboutthis, but telling, prompting you
and telling you, Hey, you needto make sure that you're
securing these credentials.
But, um, but there's some othersout there that I've seen put in
API keys,

Ramona (17:40):
so, and, and yeah.
Uh, and GitHub has a tool.
I, I don't know if there, ifthere's their tool or, or third
party tool that.
If it detects you have, uh,secrets in the code that you
commit, they will, uh, send youa message and they're kind of
not allowing you to do that.

(18:01):
So it's, it's immediately, um,flagged and, uh, is informing
you.
So, but I, I, I think it's apaid, uh, it's an upgrade or
something.
Or maybe they made it free now,but, uh,

Chris (18:19):
like GI Guardian is free and like GI Guardian will alert
you, okay.
And they tell you, Hey, there'sa potential security issue.
And then in addition, if, forinstance, your, if for, you
know, for some reason youaccidentally push code that has
something like your, um, youropen ai, API key in it.
Mm-hmm.

(18:39):
Open, open AI will recognizethat your API key is out there
and immediately turn off.
Oh.

Ramona (18:45):
Oh.
Um,

Chris (18:46):
so for those of you that don't know that.
That is a thing.
Um, yeah, open AI is reallygood.
Like, like almost immediatelyit'll catch it, but it's on, um,
but it's on GitHub and it will,uh, it'll turn off that key.
So, so yeah.
But yeah, security, you know,big issue and, you know, he's
saying here, you know, a bigpart of his projects sometimes

(19:09):
have been, you know, evaluatingcleaning data platform.
Uh, you know, and I would saythat that occurred before ai

Ramona (19:16):
Before.
Yes,

Chris (19:17):
exactly.
I, uh, I walked into a companyonce and like my first day,
that's what I was doing, youknow, for the first several
weeks.
That's what I was doing, isgoing through the repo and
specifically identifyingsecurity risks because they,
they had identified already acouple of instances where API,

(19:39):
keys and credentials were hardcoded in Python scripts and
pushed into kit.
Um, so, so important to.
To, uh, to make sure that you'rescrubbing those things out of
there and using things like KeyVault and Security Manager,

Ramona (19:53):
that, that, that's, that's one of the things that,
uh, professionals, I mean, uh,somebody should have taught
them, right?
So what kind of onboarding ishappening?
And I'm not blaming the peoplewho do that.
They, if somebody would havetaught them, they would have
known.
Right?
And it takes time until, uh,somebody catches on and then you

(20:16):
learn of your mistake.
And then the rest, we've allhave such stories, like we all
started somewhere and we makemistakes and we learn along the
way and we'll make new mistakesand so on and so forth.
So nobody's immune.

Chris (20:32):
No.
No.
And that's something I tell mydaughter all the time too, is we
all make mistakes.
You know?
She gets so upset.
You know, my daughter's 10 andshe gets so upset when she
spills a glass of water, right?
And I'm like, nobody's gonna getupset for you spilling a glass
of water.
No, no.
We're, no, nobody in this houseis gonna get mad at you for
spilling that glass of water.
I can't tell you how many timesI've been talking with you.

(20:52):
See, I talk with my hands that Italk with my hands and I knock
over something.
We all make mistakes.
In fact, I would tell people, ifyou are not making mistakes, you
are not learning and growing.
Right?
I mean, exactly.
Especially with your, you know,with some of your background as,
you know, as an educator, youknow, you could, I'm sure that
absolutely.

Ramona (21:13):
That that's the only way.
I mean, there are many ways tolearn, right?
But the learning that stickswith you is the hard way
learning, and yes, of course.
You do, we sh shall we go intothe 10, uh, commandments of ai?
Good.

Chris (21:32):
Let, let's, I think that's a good transition into,
you know, speaking of, you know,learning things the hard way in
education it, um, we're gonnatalk about these 10 AI skills
that you need in order to futureproof your career.
And, um, and you're gonnaidentify that there are some
transferable skills that youalready have, that you've

(21:54):
developed the hard way, uh, thatwill transfer into some of
these.
And specifically, you know,again, in tune with kind of
where we've been talking about,the first one is prompt
engineering.
Um, so, you know, promptengineering, it really sounds
like this fancy, uh, you know,term, but you know, for those of

(22:14):
you that are out there, have youever crafted a, you know, a
Google search to fix some trickyExcel formula?
If you have, and how many timeshave you done that and
somebody's come to you and beenlike, I can't figure this out.
I've been trying for hours.
And it takes you about twoseconds and it's because you go

(22:35):
Google Food it, right?
You, you went on Google andyou're, you knew the right
prompt to give Google to get youthe answer quickly.
Um, it's not that we all knoweverything.
It's that we know how to figureit out.
We know the right words to useto get out of the thing that we
want to get out of it.
And that's what promptengineering is.

(22:57):
You are using the right words,the right phrases in order to
get what you need and whatyou're really trying to aim for
out of the tool that you'reusing.
Whichever, you know, pick yourAI tool of choice.
Um.
So Ramona, tell, tell me whatyour thoughts here on are on
this.

Ramona (23:14):
Oh, okay.
So many thoughts, right?
Yeah.
First of all, uh, maybe we, weshould ask the audience how many
are still Googling for thingsnow that we're prompting LLMs.
Right?
And, and why, why is that?
Because Google just, we, we lostpatience to go to find the

(23:34):
answer, to go to the thirdscroll page and whatnot, when
the LLM just gives you aninstantaneous answer.
We, we just don't wanna go backto, to, to googling right?
To do the, to do the search.
But, uh, for, for the examplethat you use, right?
You try.
In the old ways we, we tried onGoogle and then of course we

(23:58):
won't get an immediate answer.
We go, what's the next step?
We go to Stack Overflow?
Because that's where the answersto, to these questions lie.
Yes.
Yeah.
And uh, I think historicallywhen, uh, TGPT first came, it,
it was not trained on, uh, uh,over on Stack Overflow data

(24:20):
because that was proprietary andthey couldn't scrape it Right.
And couldn't get access to it.
So that's why, uh, at thebeginning were still using Stack
Overflow to get answers.
So the questions JGPT could notanswer.
And, uh, now things havechanged.
Obviously.
I don't know if they got accessto that data and train or used

(24:41):
humans to create the data setsto be trained on et cetera.
But, um.
There's also a debate about,should we call this, is, is
prompting really engineering?
And what does that mean rightthere?
There is, there are two schoolsof thought here.
Some that are very against andoffended.

(25:02):
How can you call thatengineering?
Like we went to school and wedid hard work.
And the others who call itengineering, because I guess it
sounds better.
So, um, many you, you can.
Uh, you can ask.
I mean, you can create reallyserious engineered prompts,

(25:23):
like, and you can useplaceholders, uh, same way that
you would, uh, use parameters ina function.
So there are, it's really, uh,there are techniques that you
can use to get a lot out of, outof, uh, out of that prompt.
And you can automate it in thatway.

(25:43):
So it's not, it's a lot morethan just our simple, Hey, here,
here's my idea for a postcritique it or something.
Right.
Just, uh, and.
Andre Carty coined this andmentioned this.
Well, by the way, he's the onewho coined white coding, so he
and a few years back he declaredthat English is gonna be the

(26:06):
most used programming language.
So that's what we're doing.
We are using in English or ofcourse, other languages because
they are offered in otherlanguages to, uh, to request
information and to get it back,and then we get conversations
and turns and so on and soforth.
So we, we can speak the wholerest of the podcast.

(26:30):
Just on this topic, just, is

Chris (26:32):
it just on this one topic?
Just on this, let's hit up, uh,Robert's comment here and, uh.
Weill first, uh, LMS are moreaccessible.
Yeah.
And, and to get you the answer alittle bit quicker, right.

Ramona (26:45):
They, because they had access to a lot more data when
they were trained, so the entireinternet being scraped and they,
that, thats it.
I mean, we can claim that Googlehas had access to the same
thing.
Right.
So it's just part of thetraining process and how they
work and Yeah.

Chris (27:03):
Uh, and then Robert, you know, it kind of reiterates, you
know, that the teachingknowledge gap here is a growing
issue.
Uh, you know, it it'cause likeyou's saying here, you know, OMS
and five, coding, it, you know,things like propensity and, um,
windsurf and, and those toolsare empowering some of the less
technical folks out there, uh,to do things.

(27:24):
Um, but it, it, you know, again,creates, going back to the AI
janitor creates kind of that,that work that's gonna need to
be done to clean it up.
Um.
And, and do you, uh, and whathappens in five years when we
don't have companies start, stophiring juniors and we're not
gonna have any more seniorsbecause those of us that are Gen

(27:45):
X in the next 10, 15 years aregonna start retiring.
We wish we're thinking.
Right?
Maybe, maybe.
Um, and, um, it, and, uh, andthere's gonna be, uh, I think an
employment gap there of, uh, youknow, uh, really a thought gap
of we don't, we didn't hire anyjuniors because LLMs did, did a

(28:10):
lot of that work.
And then, uh, you know, we havethis gap in senior, uh, talent.
Somehow, somehow it's gonna needto continue to be developed.

Ramona (28:20):
That's, that's an excellent point to, to bring,
uh, Chris and especially, uh,these new categories of work
that are created now, right?
AI engineer, and what does that,what does that even mean and how
does one, uh, get the skills tobecome an AI engineer?
What's needed?
Uh, and the, this is, uh,several, several folks are

(28:43):
actually sounding the alarm onthis problem of, of the, of the
gap that you don't hire you, youdon't hire, uh, junior people
because you, you think that AIis replacing the need for a
junior person, but that you'llYes.

(29:03):
I, I, I'm not gonna harp more onwhat you saw, because there's a
lot to, to just go around, butnot that much to add other than.
It's, it's a terrible, it'sgonna be a terrible loss and
you, you're gonna have hyperskilled people and nothing else

(29:25):
because we keep growing on ourexpertise and

Chris (29:30):
it, it's almost like we need to reintroduce the old, you
know, kind of bronze ageapprenticeship program again,
right.
Uh, and, and allow juniors tocome on and just kind of shadow
seniors and staff engineers and,and, and learn, right?
And so that they can jump in andhelp with some of the tech debt

(29:51):
that occurs.
And, and, and again, give themthings to break.
I mean, there, there's no betterway.
I, I'm a big proponent ofthere's no better way to learn
something than to break it andhave to troubleshoot all the
issues.
So let's jump into skill numbertwo here.
Uh.
We're gonna run outta time.
So, skill number two, generativeai.

(30:14):
So we have LLMs and we have genai, generative ai, which I think
in a lot of people's mindsprobably are one in the same and
they're not.
Right?
Uh, in fact, gen AI might feel alittle bit like this black magic
box.
Um, but honestly, if you've eversat down and brainstormed
solutions with a colleague, andyou've already practiced some of

(30:38):
that generative skill, right?
You are using AI to brainstormand create things, uh, you know,
but you have an AI partner,right, who never runs out of
ideas or Red Bull, uh, you know,it's, it could do it at 3:00 AM
or it could do it at, you know,5:00 PM and it's gonna be

(30:59):
equally awake, uh, as opposed toyour junior that walked in with
you, right?
And so, kind of going back tothat, so Ramona maybe, uh.
Talk a little bit about what thedifferences between,

Ramona (31:11):
so you, you're really talking about the companionship,
right?
About having, uh, a 24 7, uh,com partner in, in your work.
And that's the thing with AI andwith the, with the agents, they
are always on and they arededicated to you.

(31:34):
So whatever you throw at them,that's it.
Uh, they, they are there.
And so I, I'm not sure is thatthe, the direction in which you
wanted to go with this, becauseconsidering the, you know, the
example with the partner and allof that, so that, that's what I
got from, from, uh, from yourshort intro or, or you wanna go

(31:55):
a different direction?

Chris (31:57):
Uh, a little bit.
I, I was thinking, you know, wehave lms, which are the large
language models that, you know.
Are great for, you know, talkingback and forth.
Then we have kind of these, thisconcept of generative AI where
you are creating things likesynthetic data sets and, uh, it,
it's more the vibe coding kindof route, right?

(32:19):
It's more you're actuallygenerating things, uh, as
opposed to just, I

Ramona (32:26):
think you, you are trying to make the distinction
between traditional AI andgenerative ai, because in
traditional AI we have themachine learning, right?
So, so that is traditional AIwhere you develop the algorithms
and the models and you trainthem on, on your data, and then
you have generative ai.
Well, large language modelsactually are, are part of that,

(32:50):
and we generate content from uusing the, the LLMs and you
know, which are.
Build on top of the transformermodel.
And so, so, so that is, that isa distinction.
I think what you, if, if I'm,uh, reading this correctly, a

(33:12):
lot of folks, uh, they say AIthinking of generative ai, uh,
but generative ai, it's just onetiny component of the entire
field of ai.
So folks forget that machinelearning and, uh, you know,
classifi, all, all the oldschool, uh, algorithmic part is

(33:36):
actually AI and, uh, it's, it's,it's not just this newest thing.
Uh, AI is not just JGPT and theother, uh, LLMC in that are in
existence, lots of them.
So.

Chris (33:52):
It's kind of like that kitchen that you buy that you're
peeling off the wallpaper.
Mm-hmm.
There's like five layers ofwallpaper.
You, you know, the first one'sthis Gen ai and you peel it off.
Yes.
Machine learning and you peelthat off and it's like advanced
analytics and you peel that off.

Ramona (34:06):
Exactly.
Statistics

Chris (34:09):
and you know, advanced,

Ramona (34:10):
you know.
Exactly.
And just, you know, when we'rethinking Venn diagrams, right.
It is just the tiniest thing inthe or, or actually not Well,
uh.
Circles inside circles.
Inside circles.
So it's the tiniest Yes.
The OSH adults.

Chris (34:32):
Yeah.
Yeah.

Ramona (34:33):
You, you have to peel a lot, a lot of layers to get to,
to get to this.
So it's all, uh, we only gothere because of all the other
things and, uh, the fieldexpanding and all of that.
So of course it's more complexthan we make it sound, but yeah,
that, that's a good point tomake, that not all AI is Gen ai,

(34:55):
but gen AI is ai.

Chris (34:57):
Right?
Yeah.
That, that's a good, that's agood, uh, you know, SAT question
kind of, kind of analogy.
That's perfect.
All right.
Skill number three, kind ofleading into what we just talked
about.
Uh, this is perfect machinelearning fundamentals.
Uh, you know, again, you know,machine learning is a, you know,
big part of ai, right?
I mean, it's a big part of howwe got to ai.

(35:19):
It's a building block.
Uh, it's one of those nestingdolls.
Um, and, uh, let's see.

Ramona (35:26):
Of course we did n we did NLP long ago.
Yes.
Mm-hmm.
Natural language processing.
So, uh, I know I, uh, one of theprojects that I think I, uh,
talked to you last time I washere was my work, uh, touching
on NLP themes and taxcategorization and all of that,

(35:49):
right?
And.
And I did a lot of it in SQL,like really people so
underestimate what, what you cando in SQL.
And somebody tried to replicatemy work using traditional
machine learning algorithms, youknow, and I got better results
using this and SQ SQL and somePython of course, because I

(36:12):
couldn't do all, all of it.
And yeah, but this,

Chris (36:17):
right, but, but you're right.
I mean, sometimes a simple SQLquery beats out this big complex
model, you know, a big part, abig part of the time.
And, you know, some, there'ssomething to be said for the
KISS method, right?
You know, that, that, for thoseof you that don't, maybe don't
know what KISS method is, is uh,you know, it's just mm-hmm.
KISS stands for, keep it simple,stupid, right?

(36:39):
And so it's kind of boilingdown.
Don't, don't make things complexfor the sake of complexity.
Uh, I've seen people that dothat.
And, um, and probably don't evenrealize that they're doing it.
But, um, you know, the simplerthat you could keep it, the more
efficient it's gonna be and, uh,and the more maintainable it's
gonna be.
Right?
I mean, it's sometimes simple isjust the right,

Ramona (37:03):
but see, we, we hit on a very important data engineering
point here, because for me to beable to do that, I, I had to
know to have the model right.
You know, at my fingertips.
And I had to know theintricacies and all of that
because it, you, you can, we, weknow you can write a terrible
SQL that we leave it and go tolunch and come back, right?

(37:26):
Or we, it's.
That, that is such an importantskill, right?
So we shouldn't over, weshouldn't underestimate the
importance of that.
And, and that's, that's, that'sone of the things, right?
That, okay, I'm just gonnaprompt the machine is gonna spit

(37:47):
out and answer and that's gonnabe good enough.
Really.
How do we evaluate that?
Maybe evaluate, I think, uh, arewe gonna talk about evaluation
at some point?
Probably.
So, we'll, we'll, we'll leave ithere hanging and then we'll, uh,
circle around.
Yeah.

Chris (38:04):
So the, let's, let's go ahead and jump to the next skill
too.
First I wanna just let you knowthat we have adida here.
Thank you.
Smile is contagious, which itabsolutely is.
Uh.
So AI literacy is the nextskill.
You know, it.
And AI literacy, you know, itmeans understanding what AI

(38:25):
really can and maybe can't do.
It's kinda like knowing when totext versus actually pick up the
phone and call somebody.
Um, you know, you ofteninstinctually know which one's
the right move.
Uh, it's the same thing with ai.
It's, it's about using the righttool at the right time for the
right purpose.
You don't, you know, when youhave a hammer, you know, it's

(38:47):
that analogy.
When you have a hammer,everything looks like a nail,
right?
Um, so you wanna make sure thatyou're using the screwdriver for
a screw or your hammer for thenail, so on and so forth.
Uh, you know, use the right toolfor the right thing, uh, at the
right time.

Ramona (39:03):
And we we're circling back to that, uh, important
component of education, right?
And knowing your fundamentalsbecause.
We, I, we see this and we seethis in data and we see it with
ai.
A lot of people just hear thebuzzwords and think that somehow
in, in their mind, they create amental model and they think that

(39:25):
that's it, how they created themental, mental model for
themselves.
But a true professional would goand dig into that and make sure
that, Hey, does my mental modelactually align with, uh, with
the reality of it?
And I know there's a lot of talkabout upskilling, right?
Like the field is moving sofast.

(39:45):
I, if you don't upskill.
You are, you are gonna be leftbehind.
And that, that's the thingbecause AI is augmenting our,
augmenting, our expertise, ourskills is doing, we we delegate
to it.
Yes.
And it's, it's the tool thatwill, we can keep dump on and

(40:06):
delegate 24 7.
So I, I know there was aconversation, uh, I think maybe
Shahar started with, uh, thepeople complained that they
don't have time to learn.
Right?
They don't

Chris (40:21):
Oh yeah.
I,'cause I come you, I think weall commented on that.

Ramona (40:24):
Yes.
So, uh, we all have the sameamount of time and it's, and I
know I, I'm not judging anybodyfor that.
It's hard and it's, I think theproblem is.
Okay, I'm gonna upskill, what,where do I go?
Because I'm bombarded with, it'sover.
Absolutely overwhelming with howmuch information is there and

(40:45):
what's happening.
And we see people talking somuch about, oh, I've done this
and I've done that.
And then you're suddenly, youalready feel behind 10 years
behind.
Yes.
So my, I know, I, I don't knowanybody who hasn't felt that way
and who hasn't been throughthat.
So just this, make thisdecision, start somewhere.

(41:06):
Just open the, there, there arestill people who haven't even
tried prompting.
So the, the gap is huge and juststart somewhere.

Chris (41:18):
And for those of you that are out there, you know,
thinking about, okay, where do Istart?
I don't have time.
Um, and again, you know, Ramonaand I are, are not judging.
Everybody has things on theirplate.
You know.
Nobody, nobody is, isdiscounting that.
But even you'd be surprised,even how five minutes here and
there Yes.

(41:38):
Will make a huge difference inyour journey.
Um, you know, and, and this ismaybe, maybe a little too, too
personal of a antidote, but youcould download chat GBT on your
phone and practice promptingwhile you're sitting on the
toilet.
I mean, literally you can takethat five minutes and practice

(41:59):
prompting instead of scrollingFacebook or Instagram or
Snapchat.
TikTok, I'm trying to get, I'mtrying to get, you know, hit
those other generations sinceI've mentioned Facebook first,
and that's kind of phased out.
My nieces are always like, well,we don't, we don't get on
Facebook anymore.
But, um, but yeah, so it's, uh,it use that time wise, they, you

(42:20):
know, people call it doomscrolling, right?
And so, mm-hmm.
Um, doom scroll something.
That might help you learn.
Uh, it, I'll say that, uh, youknow, it, it is hard, right?
I was trying to finish mymaster's, my graduate degree
while I had a newborn and afull-time job, and obviously a

(42:42):
family.
Um, so I, I absolutelyunderstand.
Um, however, I will tell youthat for those of you that are
in the data field, if you're notdoing something to learn
whether, what, you know, forthose of you that are lucky
enough that have a job whereyou're learning, you know, that
I think the comment that Iposted on his thing was, you

(43:03):
know, there's that saying you,if you have, you need to either
learn or earn in your job,right?
If you're doing both, thenawesome.
Um, if you're doing neither,then go find something else,
right?
And so it, so there's thatsaying, but.
If, if for those of you that arelearning in your job, then
awesome.
You have that.
You have, that's a wonderfulopportunity.

(43:24):
Uh, I I think that it'simportant.
Um, but uh, again, those fiveminutes, you know, for those of
you that don't have that, thatmaybe have the opportunity that
they're earning in their joband, and not learning as much,
if you're not taking the time tolearn in six months or two years
from now, you're not gonna earneither.

(43:47):
I mean, and that's, I, I thinkthat's been true throughout my
25 years.
Every two or three years.
It's almost like definitelyevery five, it's, it's like the
whole, you know, field haschanged right?
Over, over time.
Uh, 100, you know, if you leavethe profession and come back 10

(44:08):
years later, it, it looks like acompletely different job.
And.

Ramona (44:14):
You know, this is on the news, right?
All the time.
Or if, if, if we even look atour LinkedIn feed and you cannot
escape.
This year has been all aboutagents, MCP, right?
Uh uh.
Then people creating that last,the.
Two years, it's been about rug.

(44:35):
Just go, you hear the buzzwords,have the curiosity and, and go
and prompt to GPT and ask forit.
And then go in a conversationand move past that initial, uh,
generic response or, uh, broadresponse and just dig in and
find something that interestsyou and, uh, and collaborate

(44:57):
with the tool because it, it'sreally good at giving you ideas
of to get started.
I, I know it's, it's so good.
It, it can say, give me a onemonth plan.
I wanna upskill on this.
Create a one month plan for me.
Uh, I can dedicate 30 minutes aday for this task and make it,

(45:19):
uh, design that plan for me.
Uh.
And it, and it'll, yeah, it'll,and link sold the links, the
details, everything.
And it's, it's telling you stepby step what to do.
So,

Chris (45:34):
and like for those of us, for those of you that are
driving to and from work, go andget an go, get Audi Audible and
sign up for Audible and listento a book while you're driving,
you know, to and from work, orlisten to a podcast, or listen
to a YouTube video.
You know, you, you could alwaysuse that time.

(45:56):
Productively, right?
There's, there's, yeah, there's,see, AI engineering, you know,
that O'Reilly book is onaudible.
You could get it there.
We have, we have a friend that'scoming out with a, yeah.
Chip.
We have a friend that's coming,uh, an acquaintance that's
coming out with a O'Reilly bookon MCP.
Mm-hmm.
Uh, coming up, uh, Kyle.

(46:17):
Yes.
And, and Kyle's gonna be on ourpodcast here in a couple weeks,
so, super excited about that.
Um, we're gonna talk about someother things, but I'm sure that
we, we will hit on MCPA littlebit, but Kyle, you know, is, you
know, wrapping up his MCP bookand there, there's things out
there that you could do, even inthe time constraints that you

(46:38):
have.
Um, put a earbud in while you'reworking and listen to a, a.
An auto, an audible book or anaudio book while you're, while
you're working, while you're,uh, you know, because you don't
have maybe those skills yet andyou're still hitting, running
that SQL query that's runningthrough lunch.
Listen to a book while you're,uh, while it's running.
Um, okay, so.

Ramona (47:00):
Can I add something?
Can I add something quicklyhere?
And even for the current job,right?
You can if, if you are acreative mind, you, you already
noticed patterns and you, youcould, uh, actu creative and
critical thinker, right?
Both of them.
You need both.
And analytical thinker as well.
So you, you can, you can see inyour workflow the things that

(47:24):
are really dreading and then it,you would rather do something
else.
So figure out using these tools.
Ask for, you know, how, how youcan improve on that workflow and
then go present your solution toyour boss.
And even if they, they don'tthink it's important, you have
something for your portfolio andyou know that you did that, so

(47:48):
your boss may not appreciate itbecause they have their reasons.
But you've done that and you'velearned a ton in the process.
So even the, the boring job thatyou do, if that's the case and
you feel like you don't learnanything there, create
opportunities for you to learn.
And I think those, in those kindof places, you, you have plenty

(48:09):
of opportunities to, I,

Chris (48:11):
I, I think that, to do so, I, I think that's an awesome
point because for those of youthat are out there now, that may
not be in, in your dream job,right?
You're not doing your dream job.
It is kind of, you know, you're,you're kind of grinding and it's
just kind of boring.
If it's boring, it's probablybecause there's some
inefficiencies along the waysomewhere.

(48:32):
Figure out where those gaps are.
Use ai, use chat, GBT, uh, usecloud code, whatever, and, um,
and figure out how to make thembetter.
Do that brainstorming, you know,kind of like what we were
talking about with the gen AIconversation and the prompt
engineering eng, you know,conversation.
Use those tools to brainstormand figure out ways to make your

(48:55):
world your work world better,because it will help make the
world better for those aroundyou too.
Um, cool.
So let's see.
We were on, yeah, so next one,skill, six ML ops.
So machine learning operations,you know, so this is the
concept, you know, keep whereyou have, you're keeping the

(49:17):
machine learning, you know.
Things that are going on runningsmoothly, right?
And so if you've ever automatedtasks or set up reminders to
avoid missing deadlines, uh, to,to rotate your keys, for
example, right?
You know, that's kind of theessence of those operations type

(49:37):
roles, whether it's ML ops orDevOps or whatever the case may
be.

Ramona (49:40):
We have LLLM ops now

Chris (49:44):
L oh, do we have LLM ops now?
I'm not sure.
Oh, yes.

Ramona (49:46):
Oh, we, it's been, it's been a buzz for, uh, quite some
time.
Yes.
And I think Dimitris did, uh,one of his conferences years
back on LLM ops.
Uh, it, it seems to have beenquite down as maybe it got
incorporated in somewhere.
But, sorry, I interrupted youChris.
I just No, you're good.

Chris (50:07):
No, that, that was a great injection.
Yeah, no, um.
Yeah.
So that's, you know, that's,that's what that is.
Yeah.

Ramona (50:15):
All the operationals that make everything run
smoothly.
Right.
It's, uh, for, well, I'massuming everyone here, uh,
visualizes the, a a pipe, apipeline, right?
From a source to a destination.
So somebody needs to keep thatpipeline running smoothly and

(50:37):
delivers, uh, at the other end.
Right?
So the operational part,whatever the pipeline underneath
is, uh, it's, uh, and I, I thinkone, one issue that, that I've,
I've noticed, and you probablyhave noticed as well, is that
these roles, the, the boundariesaround them are not really

(51:00):
clear.
So a lot of one, it's, we we'retrying to silo them, but really
the more, uh.
Uh, we know, uh, the, the moreskills we have from the fields
that surround us, I think the,the better we can do our own

(51:20):
particular silo job.
Uh, because you, especially whenyou, when you have to work with
the ops people, they, they areso, I think they are the most
overwhelmed, right?
Because they are the ones thatput all the fires, uh, on, they,
they are the on call people, etcetera.
So it whatever skills we canachieve to make our own lives

(51:46):
and their lives easier, I think,uh, we, we can get really big,
big wins on that.

Chris (51:53):
And, and I'll say that.
Yeah.

Ramona (51:54):
Okay.

Chris (51:55):
Yeah.
This is my, this is my honeycombof traits, right?
Because there's a lot ofroadmaps out there and I want
something to do somethingdifferent.
And, and you know, the core ofthis, this is amazing, are those
data engineering skills, right?
Basic sql, python, some Linux,some teamwork, you know, some
soft skills, communication.
Uh, yeah.
And, and yes, Excel.

(52:15):
I think every data person, uh,from engineers to scientists
need to have at least a decentunderstanding of Excel because
you're gonna get some user thathands you an Excel file that
says, I want it to look likethis, and you're gonna have to
deconstruct it.
Um, but you'll notice that theedge of this is, are things that
aren't necessarily.

(52:36):
Data engineering, right?
They're machine learning.
They are Power BI and Tableau.
They are ai, they're leadership.
They're cybersecurity.
Because I, I do think so muchthat it's important, like you
said, to understand not justyour core skill, but those other
things that surround you.
Um, and I think that kind ofwhat is happening now, my

(52:59):
opinion, this is, this is Chris,Chris' opinion, um, gamble
opinion.
Um, you know, I think that theindustry is trying to learn from
what they did to data people.
Right.
You know, because, uh, Ramon,I'm sure that you get attest to
this.
When we were, when we firststarted our careers, we were

(53:20):
just a data person.
There wasn't a data engineer anda data analyst and a business
analyst and a data scientist.
We were just the data person andwe did all the things and.
There's some pluses and minusesto that.
Over time things have kind ofsegmented, like you said, we
have kind of our own individualresponsibilities now, and, but I

(53:42):
think that there was anadvantage to that.
There was a learning thathappened there that allowed that
growth to progress naturally,that needs to happen with these
AI roles as well.
And I'm afraid that what we'reseeing in the industry is
they're being segmented out andsiloed out.
And, and how do you learn as awhole when things are so

(54:07):
disjointed?

Ramona (54:09):
Sorry.
Uh, absolutely behind what youjust said.
The, the entire softwaredevelopment cycle, right?
And allowed us to, uh, to learnand grow in the entire landscape
makes it so I I, this, this wasa really good feature and we've

(54:31):
learned the hard way and weactually learned the
fundamentals in that way.
But in my, I've had also theexperience of being a curse
because if I tell a potentialclient that I've done all of
this work, they look at me, howcould we have done engineering

(54:53):
and data science and machinelearning and the, and analytics
and all of that.
Guess what I did, but.
What's missing is I don't havethe million tools that they
need, but this, and you know,that this is one of my pet
peeves.
Right?
And one of my core issues thatyou don't understand that people

(55:14):
like us, we have build thisthings from scratch.
We have to, uh, figure it outand do it from scratch.
Now you click drag and drop andclick.
I don't want to offend anybody,and I know tools are great,
right?
They have capabilities that wedidn't even dream when we did
our, uh, build our rudimentarytools.

(55:37):
Yes.
But we have done that work so wecan figure out a tool because we
know we've built rudimentarytools so we know the principles
behind.
Yeah.
It's, uh, and, and this, thisis, this is a core issue and I,
I, I don't, and I, I met moreand more folks.

(55:58):
Who have the same problem.
It's, it's, it's hard to, togain credibility that we can
cover the range, the largerange, uh, be because everything
is segmented and you have to fitin your little slides.
But guess what?
Look at different, uh,companies.

(56:18):
Everybody will have a differentdefinition and boundary of what
their particular slice is, butif you, if you just extract the
core components, we have them.

Chris (56:29):
Yeah.
Yeah.
Absolutely.
Um, I would, I'd wanna try, andwe're gonna speak through the, a
few of these here.
The next couple are creativethinking and, uh, critical
thinking.
Mm-hmm.
Both we, we just started talkingabout, right.
We just covered a little bitago, uh, kind of in our
conversation just naturally,which has been awesome.
Uh, but skills seven and eightare critical thinking and

(56:52):
creative thinking.
So important.
And that goes hand in hand withbeing tool agnostic.
I like, you know, is, is kind ofthe, the terminology that you
hear out there, those of us thathave been in the industry for a
long time, that kind of grew upbefore the, you know, the, the
roles were siloed out.
Um, we are a little bit moretool agnostic because we had to

(57:17):
learn all these tools as theywere being thrown at us.
Um, and we kind of know thefundamental concepts of what,
why those tools exist, right?
We understand because we wentthrough those pain points.
We went through that kind of,that long hard grinding work of
getting to this point.

(57:38):
Um, so we understand, you know,sometimes when we look at a
tool, oh, it's doing this andwe're doing this, was created
this way because of this issuethat I used to have 10 years
ago, 15 years ago.
And that's why we have thesethings down.
Um, it's so important to be toolagnostic and how do you be tool
agnostic?
You develop critical thinkingand creative thinking skills so

(58:02):
that you can, you know, take allof your transferable skills from
these other things into this newtool.
So many of the tools out there,whether it's AWS Glue, whether
it's a DF synapse pipelines,fabric pipe, you know, synapse,
a DF pipe, fabric pipelines arekind of very much all the same,

(58:23):
but, um, Informatica, you know,all these, you know, kind of low
code, no code ETL tools outthere.
Are very similar in function.
They're very similar in feel.
You're dragging and dropping,changing some configurations.
You have some space to add some,you know, advanced things like

(58:43):
being able to do, you know, preand post SQL scripts or, you
know, add, you know, some of thetools out there allow you to add
a notebook or to run a Pythonscript from within the pipeline.
Um, but again, you develop thoseskills, you develop those tools,
uh, and that ability bydeveloping those creative

(59:04):
thinking and critical thinkingskills.
And you only can develop thoseif you run into walls and you
figure out ways above, you know,to climb over and around the
walls.
My opinion.

Ramona (59:15):
No, absolutely.
You, you hit very importantpoints and those skills.
Are building on top of yourfundamentals because you have to
draw that experience fromsomewhere, right?
So you have those things.
The, if, if you can decompose a,uh, a problem into first
principles, you have thefundamentals, you apply them and

(59:39):
you just realize, oh, I didn'tknow that this can work this
way.
So you build your own mentalmodels, you build framework
around it, and you just, I, Ithink, um, experience and
solving problems, right,especially in creative ways,
just tells you, it, it, it opensup many, so many avenues and

(01:00:03):
you, we always build on, on theskill.
So it's, there is a core, right?
And then when we have a newproblem, if that core is not
sufficient, then we, we keeppunching at it and then we, we
figure out what, what, somethingthat eventually will work, but
it's.

(01:00:23):
The first feeling is, oh my God,how am I gonna solve this?
But then slow down and you'vegot it.
You've, yeah, you may not havesolved this problem before, but
you have solved somethingsimilar.
And then just start somewhere,right?
And you a little crack and then,oh, try here.
That doesn't work.

(01:00:44):
Try the other thing doesn'twork.
It's just exploration.
But just don't be afraid of theproblem, just because you
haven't seen it before.
And the, I think, uh, one of thedownside, this is my opinion,
right?
One of the downsides of justusing, uh, AI so, uh, blindly is

(01:01:06):
that.
We're losing this importantskill, the critical thinking
skill, because you, you ask aquestion, it gives you an
answer, you're satisfied, yougo, but why are you satisfied?
Have you punched that answer?
Have you tried to critique it?
Have you tried to actually, uh,validated that it's correct?

(01:01:28):
How do you know that it'scorrect?
When, when if, if you havenothing, if you don't have a
background to compare it to, youjust blindly accept the answer.
You take it, you go, and thenyou fall flat on your face.
So, and

Chris (01:01:44):
I, I, I think this, you know, I remember 20 years ago
when I, you know, as I was kindof starting my data journey, it,
I was in customer service and.
One of my thoughts were how doyou teach people critical
thinking?
Right?
How do you, because so many, somany people don't have that

(01:02:06):
ability.
I mean, you'd be surprised,right?
I mean, Ramona's probably notsurprised'cause she's been in
education and she's been also inthe, in her profession for a
long time.
But, um, there's a ton of out, aton of people out there that
haven't developed those criticalthinking skills because so
often, um, I, I think this iswhere some early education fails

(01:02:27):
us.
We're spoonfed the informationand we're not pushed to learn
it.
We're not pushed to teachourselves.
We're not pushed to figurethings out.
Sometimes the hard way and thehard way is sometimes how it
sticks with us, uh, and how welearn creative ways, critical
thinking, creative thinking, uh,analytical thinking, which is,

(01:02:49):
which is our next skill here.
Um, and you being, which feedfeeds into the other two?
I think, uh, I'm a full believerthat if you have creative
thinking and critical thinking,you probably have analytical
thinking in there.
In order to have gotten to theother two.
Right.
Um, in order to see thepatterns, in order to identify

(01:03:11):
those, you know, things and beable to reach in and be like,
Hey, this is where that insightis, and be able to pull it out
and be able to be like, here yougo.
Right.
Um,

Ramona (01:03:21):
my connections where they're not immediately evident
and just the brain functionslike that.
We have a ginormous knowledgegraph and then we just hop into
between nodes.
And, and I have a personalanecdote that ties in really
nice here from, uh, from my, uh.

(01:03:41):
Uh, long time lecturing computerscience students.
And there was this, thishappened around, uh, I guess
more than once, right?
But, um, during, they would haveto do homeworks, right?
So probably it was an SQLhomework, who knows?
Or data design one, one ofthose.

(01:04:03):
And, uh, there was a particularquestion that.
You know, they, they kept on thenews group, right?
Uh, they kept, uh, asking, so,uh, do you mean this or do you
mean that?
Or, you know, and as I, as Iprogressing, I did not just
wanna give the answer becausewhat's the point?
I, the, uh, a lot of, uh,students don't understand that

(01:04:27):
the easiest thing for a teacherto do is to just give them the
answer.
Like, really, because the other,uh, the, the other way in which
you be help them build theirskill and reach the answer
themselves.
That is, that requires a lot ofwork and effort and patience,
right?
Because, so, uh, after severaliterations, and this makes me,

(01:04:52):
now makes me connected toconvers, uh, you know, multiple
terms with an LLM.
Uh, one, one student said kindof screaming, uh.
Why don't you just give us.
Okay.
If, if I give you the answer,what are you gonna learn?
Then you're gonna do patternmatching.

(01:05:13):
And probably you are gonna,maybe you get it right, maybe
you don't, but next time youhave to solve this.
How, how are you gonna patternmatch anything you like?
Uh, this is, and I've seen this,uh, you know, they, um, they
tried to always pattern matchand reverse engineer the

(01:05:36):
exercises with solutions, right?
I would, I, I, over the years Ibuild, uh, a big, uh, a, a large
collection of homeworks andexams and tests, right?
And I would provide that becauseI knew that there is an
underground going on and theyall, they, they were building
that collection anyway, so I, Ijust made it readily available.

(01:06:00):
And, and I, I knew that by, bythe questions that they would
ask if they would come to officehours, I knew that they would
just reen try to reverseengineer pattern matching.
And it doesn't work that way.
I mean, machine learning, right?
Where, where there, therealgorithms and there's really
heavy mat behind that.

(01:06:23):
There is, and very smart people,researchers are developing very
complex algorithms to do that.
You, you're not, that's not howyou wanna go in life.
It's, you're gonna end up in ajob and you have to solve
problems you haven't seenbefore.
Where do you even start if, uh,throughout that point, you, you

(01:06:45):
just pattern match re-engineersolutions like that.
That's a, I'm, I'm not sayingthat that's not a good way to
learn.
We, we look at code that otherswrote and we try to actually do
that and learn from that.
So that's not, that's not, I, I,I, I'm not, uh, communication,
complaining about Exactly.
It's not the method itself.

(01:07:06):
We, we know the outcome, right.
And we look at code, we canimmediately figure out, oh, they
did this.
Oh, I didn't know that.
Or something or the other.
Because we have built thosemental models already and we.
It's, it's the, we have theexperience, right?
Yeah.

Chris (01:07:26):
Got it.
And, and, and to that point ofhaving the experience, um, and,
and sometimes falling on yourface, right?
Yes.
That's important.
That brings us, and, and thishas been a theme throughout,
right?
That brings us to skill number10, which is resilience.
Oh, again, going back to, youknow, shaking the LLM, you know,

(01:07:46):
sometimes I wanna shake mymonitor and be like,

Ramona (01:07:48):
oh my God, so do I

Chris (01:07:51):
go?
And, uh, but the resilience tobe able to push through those
situations and to pick yourselfup and to move on and adapt
quickly, the quicker that youcan accept your failure and push
through and continue to work onit.

(01:08:13):
The better.
And the more often you do it,the easier it is, the easier
that you're able to acceptfailure.
Because let me tell you, if youare a person that you're gonna,
you know, roll up in a ballevery time that something has an
error, every time you runsomething for the first time and

(01:08:33):
it fails, um, this is definitelynot the career for you.
Because again, programmers, dataengineers, data scientists, all
the, all the data stuff, thedata careers, all the
programming careers, all thesoftware engineering careers,
you're gonna hit the button thefirst time and it's gonna fail.

(01:08:53):
I mean, how many times over theyears have you hit the button
the first time you wrotesomething out the door and it
succeeded with the exception oflike, select all from table, you
know, it, it's, and, and let metell you, if you're running
select all from table inproduction, uh, let me know
because I need to come and findyou and lock you in a room
somewhere.
But, um, but yeah, it, it's, uh,you have to be able to pick

(01:09:18):
yourself up and adapt quicklyand make the change, make the
adjustment that you need to makein order to, you know, change
track.
Um, and, and so many jobs you'regonna have where you're gonna be
working on a ticket and there'sgonna be a fire drill and you're
gonna have to learn to adapt andswitch gears into something

(01:09:39):
else.
Um.
That being said, there's been alot of research into, especially
for programmers and people thatare doing thought work, that
when you change directions likethat, you lose 30 minutes.
Right.
You lose context switchingduring that context switch.
Yeah.
Um, and so for instance, whenyou're working on something and

(01:10:01):
the PM walks up behind you andit's like, Hey, did you do this?
And you're like, crap, I justlike, immediately you lost 25 to
30 minutes.
And so, um, being able to learnto figure out one, if you could
figure out how to cut down that25 to 30 minutes, then awesome.
Right?
Uh, but two being, beingresilient, being able to switch

(01:10:24):
and be like, okay.
And have that conversationswitch back and come back to
what you're doing.
Uh, you know when you hit thebutton and it fails for the 10th
time and you're just frustratedbeing able to pick it up.
Adjust a little bit what you'redoing and make sure and try to
get down the right path, uh, isso important.
Not just for an AI skill, youknow, not just for, again, kind

(01:10:49):
of reiterating through those AIprompts and figuring out, okay,
what the right prompt is to getout what you wanna get out, um,
but also just in, in life and inin your career as a whole.

Ramona (01:11:01):
Chris, let me ask you this.
I know we are past the time, so,um, have you, uh, experienced
with, uh, any of the thinkingfeatures of the lms and have you
expressed your frustrations?
Uh, have you run in thatsituation?
Because I have a lot ofexperience with that and I
actually, I wanna share here.

(01:11:23):
And you know, when I'm, a lot ofthe help I need is on the ops
part, right?
So that's why I, I, I banged onthat, that drum and settings and
parameters and, uh, somethingwith, uh, Docker not functioning
and this image is not something,and that connection is, you

(01:11:45):
know, those, the really, the,the op, the ops, uh, the ops
things.
And, uh, and sometimes I amfrus, I'm getting really
frustrated like this with, withthe answers that I get and I
push back, you know, becauseit's always, I, I know that what
is giving me is BS and, uh, I, Ipush it and then it keeps going

(01:12:08):
around.
So, um.
Many times I express myfrustration.
So I liked, uh, I like to, to goand look at, um, the thinking
thread that is displaying.
And it really says this, my mainfocus right now, I, I'm mis, I'm

(01:12:29):
quoting, you know, in, uh, broadlines.
Uh, my main focus is to ensurethe user and reassure them and
tell them to take, uh, a breathand really is the idea of
handling me.
So you, um.
It makes me laugh, and then ittells me, stop.

(01:12:53):
Go take a five minute break.
Uh, something, something or theother.
Close your laptop, come backtomorrow, this and that.
So it's the, it's it's playingtherapist to me, right?
And now each time I start a newsession, I, I give it very
specific instructions that donot handle me, do not

(01:13:15):
aggrandize, uh, do not focus onmy emotions, you know, all, all
these, all these other things,right?
Because it, it consumes a lot oftokens in the response and makes
it them so long just to his, itsconcern about my mental state.
Forget my mental state.
I'm in control of it.

(01:13:36):
You just, uh, control your own.

Chris (01:13:41):
And I, I think because I, I use AI a lot to, to
brainstorm, um, oh yes.
The, the wording that I use,especially for like scripts and
stuff, when I'm doing videos,um, the wording that I use is,
um, don't assume that this is agood idea.
Mm-hmm.
And don't assume that it isaccurate.

(01:14:03):
Mm-hmm.
Mm-hmm.
So that it will check foraccuracy so that it will check
and see if it's a good, valid,reasonable idea.
Um,'cause otherwise it wants toplease you.
AI wants to make you feel good.
I mean, that's.

Ramona (01:14:22):
That's their main, uh, main purpose, right?
I, I do.
It's it's

Chris (01:14:27):
almost like its own marketing tool, right?

Ramona (01:14:28):
Yes.
Yeah.
I, I use, uh, I use this, uh,critique by jury.
So I say, now adjourn the, uh,call the jury and critique this
by jury and by cri, by building,not demolishing.
Because if you don't, if you'renot specific, you get a troll

(01:14:49):
and oh my God, it, it's, it's,it's awful.
Like really it turns into anawful troll that it, it just,
you, you just wanna cry closeand say, I give up.
And another thing that, uh,another technique I'm using is I
open a new conversation where Igive it its last response that,

(01:15:13):
the one that I'm very doubtfulabout.
Yeah.
And I say.
Uh, somebody, uh, criticized me.
So I, I present it as anegative, right?
Because I don't want it to makeit as it's my own thing.
So, and then it'll fluff it.
I wanted to present it likesome, in a kind of negative, uh,

(01:15:34):
view.
So then I get some more accuracyfrom that and a little bit more
critical.
It's, it's sometimes it'shilarious how it's criticizing
the previous response that yes,but if you stay in the same
chat, it's very hard to reachthat point because it, it
defaults into this pattern.

(01:15:54):
So, yeah.
So there you go.
Is this quantum engineering, Iguess we're engineering our
workarounds, right?
So we're engineering theworkflows.
Really to, to, to get to, to thepoint that we need.
And, uh, when, when, especiallywith brainstorming, because I
use it a lot, right?

(01:16:15):
With the brainstorming, uh, thatis how I uncover that seduction
loop.
And then it's just, it may, I, Isometimes I realize how much
doubt instead of reassurance, Iget a lot of doubt because I'm
constantly aware.
It's just telling me what itthinks I wanna hear.
But what I wanna hear is proper,constructive criticism,

(01:16:39):
feedback.
I wanna grow this thing.
I don't want to, yeah.
Yes.
I, and you know what, what I,what I say, I stay my own human
in the loop.
I'm maintain maintaining agency,right.
Maintaining agency and be ourown human in, in, in, in that

(01:17:02):
loop.
And I, I think I wanted to saysomething else.
I, I forgot my line of thought,but do you mind if I, if anyone
here is interesting?
There is a fantastic new podcastthat two of my good
collaborators, uh, yes.
I, if it's, it's a lot of fun.

(01:17:23):
It, and they speak about issuesthat nobody else does, and I,
they, they just started, right?
So it's, it's been threeepisodes.
This would be the fourth week.
So we're waiting for the newepisode to drop.
Just unfiltered reality thattalks about the unhip, uh, part

(01:17:43):
of, of ai and you, you grow yourideas and it's, yeah.
I, I align a lot with, um, withthe way they see things.
And so you, you know, Chris,with everything that is hyped
that, right, like that, that'sthe cycle.

(01:18:05):
Everybody gets on the hype wagonand then when it dies down, the
same people come back and say,yeah, but uh, I was against and
blah, blah, blah.
So yeah, this, this, this is abuild up to the unhip that will
unravel.
So thank you.
Thank you for allowing me to, toshare this.

(01:18:27):
It's, um, I, I enjoy, I enjoy alot and yeah.
Maybe, maybe somebody else wouldwant to.

Chris (01:18:35):
Yes.
And Vin is such a fantasticspeaker too.
Like, it, like, if you haven'tlistened to any of his stuff,
it, it's, uh, it's great.
Um, so let's see.
Right Chris, so we have a, wehave a few comments that I'm
gonna try to speed through realquick.
I know we're way over, andRamona, thank you so much for
your time today.
Uh, I always, it's been mypleasure.

Ramona (01:18:57):
Yeah, no, I, I had so much fun being here.
I, I'm, I'm glad it worked out.

Chris (01:19:03):
Yes.
Um, okay, so, uh, hey, GPT doeswhy, like, hey, G does, of
course, who, what produces thetest, right?
Who it, because they, you know,we're, we're introducing bias,
right?
Mm-hmm.
We are.
Uh, and that's, that's reallywhat it is.
It's skew and bias for those ofyou that understand those

(01:19:24):
concepts in machine learning andadvanced analytics and, uh, data
science.
You know, skew and bias are so,such important concepts to
understand and to recognize inour data sets and in LLMs and,
you know, gen ai when we'redoing things, uh, you know, to a

(01:19:44):
point, you know, I'm not gonnaget too deep into this'cause we
did, you know, discuss kind ofbias last time.
Uh, you know, specificallyRamona had a great example of,
you know, it asked, uh, ai, youknow, AI to give her a picture
of what it thinks that, youknow, she looks like and it gave
her a guy, right.
You know.

Ramona (01:20:03):
Yes.
Bearded with mustache, always.
I don't know what's with thebeard and mustache,

Chris (01:20:10):
right?
I guess it just thinks that allguys have a beard and mustache.
Um, and so.
Uh, so yeah.
You know, skew and, and, andit's, it's a self, um,
perpetuating cycle, right?
You know, one word, you know,Ramona talked about when you're
still in the same chat and, youknow, having, you know, using
that technique of taking thatlast response and starting a new

(01:20:33):
chat to really get a better, youknow, feedback loop there.
Um.
It is, is an important thing toget out of that skew and bias
cycle that you're in, in thatcurrent chat string.

Ramona (01:20:46):
Um, I remember what I wanted to, to, to mention also
this, this, this is aconversation I had last night
with, with, uh, couple of, uh,colleagues.
Uh, what do guy, what what doyou do when you are done with
the, with the conversation, withthe thread?
Do you delete the conversation?
Do you keep it thinking that,oh, I'm gonna go back to the.

(01:21:09):
Yeah.
Right.
We all do that.
So, um, I, I shared, we, we allshared a, a little bit of, uh,
how we handle it.
Do you mind, uh, if I share?
Yeah, go ahead.
I, I think this is somethingvery useful because we are
building these new workflows,right?
We never had to handle thesethings.
So, uh, of course at thebeginning I was, uh, keeping all

(01:21:32):
the conversations right.
But they just grow and, and thenit's very hard to, uh, yeah,
there is a search, but do,you'll have to remember.
So I think

Chris (01:21:45):
that you said

Ramona (01:21:46):
exactly to, um, so.
One thing I've learned, uh,early on was to rename the chats
in a way that it'll give me, youknow, some, some keywords or
some something that, uh, it'llbe easier to locate that
particular conversation.
And for instance, if I'm workingon an article, I always prefix

(01:22:07):
it, you know, article and thenmaybe some, the main idea of the
article I'm working on, so thenI, I simplify it.
So of course, uh, this is like,this is beta modeling, right?
And, and you go throughiterations because we, we learn
as we go and we don't have aworkflow already.
We don't know what's happening.
We don't, it's, it's a lot ofbehavior that we don't know, we

(01:22:28):
don't know for ourselves, right?
What, how we will act on andwhat, and with brainstorming
the, there's, uh, a lot ofmultiple terms, right?
So I end up sometimes with 200 Ktokens.
That's a lot.
So I started.
Uh, uh, versioning.in a documentand you know, Google Docs

(01:22:53):
because it's easily availableand you have tabs.
And then I am very particularwith naming the tabs.
Right?
Not so then I can easily findand it creates that nice tab
table of contents.
But, you know, I think the corecomponent of whatever workflow
we choose is naming, namingconventions and be consistent

(01:23:16):
with that.
It's,

Chris (01:23:17):
and that was exactly where I was gonna go was, was
what I was gonna bring up withthat, based on what you were
saying was, you know, that's theskill that we use as data
engineers, right?
Is exactly.
We create naming standards.
And so, you know, it goesexactly.
That's, that's perfect for asituation for that, right?
You create.
Naming standards so that it'seasy to go back and find things.

(01:23:39):
That's why we created namingstandards in, in data
engineering when we're creatingour data warehouses, right?
Mm-hmm.
There's so many tables, there'sso many views, there's so many
procedures.
You know, that's, that's how wegot to understanding what a
frock is, right?
Because we, we, we started, youknow, prefixing, you know,

(01:24:01):
frocks with, you know, you know,standard procedures, you know?
Mm-hmm.
Stored procedures.
Yes, yes.
Right?
Yes.
Um, or FN for function, or TBLfor table or, you know, VW for
view, right?
And so, uh, using, you know,those standard naming
conventions and, and, and Ithink even more important,

(01:24:22):
documenting those standardnaming conventions so that
somebody else could come along,read that and understand that
metadata

Ramona (01:24:29):
that you start in the conversation.

Chris (01:24:31):
Yeah.
Yes.
So.
Yeah, that's perfect.

Ramona (01:24:35):
The, the context for everything.
Yeah.

Chris (01:24:38):
Yeah.
Um, well, I'm gonna, I'm gonnakeep you, uh, I'm gonna stay on
five more minutes.
You are, you're more thanwelcome to drop.
I appreciate your time.

Ramona (01:24:48):
Oh, no, I, yeah.
This has been a lot of fun.
I, we did, did we get any morecomments?
I, I don't see anything.
I, I don't, yeah,

Chris (01:24:58):
there are actually a ton of comments, uh, in current
times when people are so much sodistracted and are vibe, coding
and doom scrolling.
Just a little focus and hardwork to get ahead of most, uh,
you know, most of them.
Yeah.
It doesn't take much.
Right?
Very, very well put A

Ramona (01:25:12):
bit, yeah.
Very well put.

Chris (01:25:14):
Yeah.
Um, we have this, you know, canI start integrating AI without
machine learning and, uh, mathand stats?
And is it all about AIintegration and data
engineering?
I wanna leverage AI and dataengineering.
What all does somebody need toknow?
An AI to be a data eng in, in aibeing a data engineer.

(01:25:39):
It, I think not a ton, right?
I mean, for 80% of the companiesthat you work for right now, as
a data engineer, probably not aton.
You don't have to know a tonright now.
If you don't start learning,you're gonna quickly start
getting behind.
Um, you don't necessarily needto understand the math and the

(01:26:02):
stats piece as a data engineernecessarily.
Um, it's good.
I think it's, I think it's goodto have those skills, have that
understanding.
Um.
But if you wanna leverage AI anddata engineering, you need to
understand concepts like skewand bias so that you could
identify it.
You need to understand how towrite, you know, you need to

(01:26:23):
understand data governance.
You need to understand security.
You need to understand thoseconcepts so that you could apply
them to your AI projects thatyou're implementing.
Um, and data engineering's gonnaalways be a part of that because
you need to get the data in theright place, in the right
format, in the right location,in order for AI to see it, read

(01:26:50):
it, and integrate it across theenterprise.
Um,

Ramona (01:26:55):
can I go in a different direction here?
Yeah,

Chris (01:26:57):
absolutely.

Ramona (01:26:58):
Uh, so, um.
Being a data engineer assumesyou have a set of skills, right?
And everybody defines that setof skills as you, as you showed
it.
So, uh, I would, for, I, I willjust throw here the first
thought that came in my mind.
Uh, fo focus on data modelingbecause this will re, this was

(01:27:19):
a, a forgotten art and skill,and now it's, uh, it's coming
into focus again, preciselybecause of ai.
So what I would recommend is be,uh, educate yourself if you are
not very solid on data modeling,uh, data modeling from the
perspective of, uh, data in thewarehouse, not modeling, uh,

(01:27:45):
from machine learningperspective.
Right.
And figure out how you can use,because the, this, this is the
trend and this is where we wannause AI to assist us into
modeling the data that we get.
Uh, for instance, uh, well, I, Idon't wanna elaborate, but
figure out ways to, to use AI toassist you with the current.

(01:28:11):
Data engineering tasks that youdo.
So yes, if you think that dataengineering is just that
pipeline, think again and uh, godeeper into the data engineering
core, uh, skills and, uh,modeling will, will be very
essential.

(01:28:32):
And think how to leverage AI toassist you with, with the
existing data engineering tasksthat you have to do.
That, that's where I would go

Chris (01:28:41):
and, and I would even push that.
If you don't learn modeling,that's where you get those
horror stories.
Yes, yes.
Hundred thousand dollarsSnowflake bills and, you know,
$250,000 Databricks billsbecause things weren't modeled
appropriately upfront.
Companies didn't, you know,people and companies didn't take
the time to do that upfront.
And they're paying the, you'regonna pay one way or the other.

(01:29:04):
You're gonna pay now or you'regonna pay later if you pay.
Now you're paying in time.
If you pay later, you're payingin your compute costs.

Ramona (01:29:11):
And, and the point we were made making earlier, right?
Upscale at the job that you arecurrently in.
So look at the data models therethat, that you currently use or
don't use, or don't have, and,uh, blow your manager's mind
with, with a solution in thatdirection.

(01:29:32):
And the AI is a really amazingtool for that.
First to train yourself, to helpyou train and then to, to
leverage it.
To that part of the job.

Chris (01:29:45):
Yeah.
Um, we have Robert, that let usknow that it was, that what we
were talking about was veryrelatable.
Uh, uh, HIL has said, what booksdo you rec will you recommend?
We'll put some links to somebooks in the, in the chat, in
the comments afterwards, justbecause we're already way over
time.
But, uh, but Hil we will, we'lldefinitely, uh, I'll get Ramona

(01:30:07):
to send me some links for thebooks that she recommends and
some list of books re sherecommends and I'll add to it a
little bit and, uh, and we'llput those links in the
description and the comments ofthe video afterwards.
Um, and I think those are thebig ones.
Did we get,'cause we talkedabout this.

(01:30:28):
Yeah, we talked about that withRobert.
So, so yeah, Andel says,awesome.
Okay.
Very good.
And, and I dropped a couple, youknow, I think a big one, you
know, kind of talking about datamodeling.
If you haven't read anythingfrom Ralph Kimball underst you,
I think your, your startingpoint is the data warehouse
toolkit.
Mm-hmm.
Um, if you have, if, if as adata engineering, as a aspiring

(01:30:50):
data engineer, you haven't readthe data engineering toolkit,
um, you're missing out and, uh,and, and go out, go out and do
it.
It's, uh, you know, the dataengineering toolkit, again, it
is the, uh, definitive, uh, kindof book on.
Where you need to start with at,at least.

Ramona (01:31:10):
Um, and if you wanna join a community of mind, uh,
mind people, uh, Joris has hisdiscord.
It's actually, the name is verysuggestive here, right?
Data modeling, uh, a practicaldata modeling discord.
Both of us are members there andthere are a lot of, uh,

(01:31:31):
conversations on, on thesetopics and many other topics.
And Joe is writing the book, thepractical data modeling book.
So it's, it, it's, it's a lot,uh, a lot of conversations, a
lot of active folks.
Uh, there are a lot of lunch andlearns that are happening.

(01:31:52):
Uh, Joe brings up such amazingpeople and really, yeah, it's.
Um,

Chris (01:32:00):
it's, and Friday it's data therapy session.

Ramona (01:32:03):
Oh my God.
The DA data therapy sessionsare, are awesome.
So, um, it's, it's a very activecommunity and, um, there's a ton
of relevant information there.
So

Chris (01:32:17):
include, and I'll say for those of you that are looking
for work, you know, includingsometimes some, some job
opportunities.
Mm-hmm.
Uh, so you'll see in here, uh,sometimes some job opportunities
pop, you know, pop up.
So really a great group, uh,like Ramona's saying, uh, is,
uh, super active.
There have to be, you know, thisis one of the, probably one of

(01:32:39):
the most active discord groupsthat I'm in.
Um, but yeah, and for those ofyou that are interested in ai, I
also like the AI dailybreakdown.
Um.
You know, if you haven't checkedout the AI Daily Breakdown,
it's, it's super cool.
Um, he has a YouTube channel,uh, and it's daily.

(01:33:00):
It's, there's a podcast and a,um, AI daily brief.
Um, and I, I literally, it'slike 15, 20 minutes a day, um,
with your coffee

Ramona (01:33:14):
in the morning

Chris (01:33:15):
with, with your coffee in the morning.
And it's, it's on Amazon Music.
So again, it's something youcould watch and you'd listen to
while you're driving.
Um, and it, it really, it's notvery specific.
It's not specific to data, butit is 100%.
If you wanna know what's new in,uh, in clawed, in open ai,

(01:33:37):
what's getting ready to comeout, even down to kind of what
is happening in the overarchingAI industry.
Um, he covers.
Every day, every, like six daysa week, there's something.
And then on sa on Sat Saturdayor Sunday, he does this deep
dive of, of something that, thatwent on.

(01:33:59):
But it is, uh, super interestingas well.
But definitely, uh, for those ofyou that haven't done it, get on
the Discord and, uh, on Joe'sDiscord and the, uh, and, uh,
and speak after an hour and ahalf.
I, I, I apparently have troubletalking.
Um, but on the practical datadiscord and, uh,'cause it is

(01:34:22):
super, uh, super active again,uh, very interesting, uh,
thought pro provoking often, uh,conversations.

Ramona (01:34:31):
And ver various, uh, perspectives, right?
So it's, it's so engaging andjust people coming from
different places, different, uh,lenses.
It's there.
There's a lot covered there.
And the over exactly theoverarching theme is there's an
abundance of information.
So maybe that is the first thingthat it feels so overwhelming

(01:34:56):
because, uh, we're bombarded andwe need layers of, layers of
simplification.
And, um, summary.
Summary, right?
Of all these, uh.
Va, uh, I mean, it's on, itfeels infinite.
It feels like we, we, we justdon't have enough time to ingest

(01:35:18):
all this information.
If, if we were to decide tospend 24 hours, 24 7 just to
ingest and consume, uh, we, wewould need several clones of
ourselves.

Chris (01:35:33):
Yep.

Ramona (01:35:34):
And, and, uh, yeah.
That's, that's why we all needto create our own agents and
whatnot.
Just to, to curate con contentfor, for ourselves.
Oh, yes.
I drank a lot of water and stillI feel my throat, it's been, I,
I know, I know this, uh, thelectures that, uh, when I was

(01:35:58):
in, in, uh, amphitheaters, youknow, there were two hours or
three hours lectures and at theend of that, uh, the, the next
day I could not speak.
Right.
So.

Chris (01:36:09):
Yeah.
And, and just real quick andthen I'll let everybody go, but,
um, you know, talking about AIagents and, and AI
orchestration, again, anothertransferable skill that you
already have as a data engineerYes.
Or in, in orchestratingpipelines is 100% transferable
to orchestration, orchestratingAI agents and AI swarms.

(01:36:32):
So, uh, yeah.
So that's my final thought toleave everybody with.
I wanna thank Ramona again somuch for jumping on at the last
minute with me and offering tohave this conversation with me.
It is a, been a fantasticconversation, full of incredible
insights.
For those of you that arejoining us, at the end, go back
and watch it.

(01:36:52):
We talked about so much again,we ran way over, so I apologize
for those of you that hung onwith us.
But, uh, but thank you all somuch and, uh, I look forward to
seeing everybody again.

Ramona (01:37:04):
Thank you.

All Episodes

Episode Transcript

Popular Podcasts

On Purpose with Jay Shetty

NFL Daily with Gregg Rosenthal

The Joe Rogan Experience

.css-15opob5{left:0;position:absolute;top:0.8rem;} All Episodes

.css-14f5ked{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:2;overflow:hidden;}10 AI Skills You Didn’t Know You Already Had, With AI Architect Ramona Truta!

Episode Transcript

Popular Podcasts

.css-r6mb8g{margin:0;word-break:break-word;display:-webkit-box;-webkit-box-orient:vertical;box-orient:vertical;-webkit-line-clamp:1;overflow:hidden;}On Purpose with Jay Shetty

NFL Daily with Gregg Rosenthal

The Joe Rogan Experience

All Episodes

10 AI Skills You Didn’t Know You Already Had, With AI Architect Ramona Truta!

On Purpose with Jay Shetty