Category: Tech Conference
Tags: AIBusinessDatabricksSummitTechnology
Entities: AI SummitConfluentDatabricksDelta Live TablesJoseLakeFlowSeb
00:00
[Music] [Applause]
00:17
I have another wonderful person that I would like to introduce you to this afternoon and that is one of our most esteemed MVPs and that is Jose. Welcome.
It's honored to have you here this afternoon. You've had an incredibly busy week.
How how have things been for
00:33
you this day at AI Summit? Oh, I think it's been a wonderful time.
I've uh enjoyed interacting with a lot of different folks. Uh I personally get to interact with a lot of different folks from the product team.
So lots of great conversations, lots of excitement
00:51
and not just fluffy excitement but there was a lot of like hey we can go back to our business after after this data bricks data and AI summit and actually start implementing some of the things like Genie dashboards and even LakeFlow. So I I I I've personally have enjoyed it
01:08
myself. I've enjoyed the content myself and that's that's a hard thing for me to do.
So yeah. So you are known for having kind of like a very specific take on things.
So go on give us your honest opinions on the keynotes. What did you like?
Where did you want to see a bit more? Maybe
01:23
I think there was overall a very solid balance between the catchy and trendy, the AI, the agentic as well as the data engineering fundamentals which everyone likes to talk about how important data engineering fundamentals uh is and
01:40
really I think the keynotes and a lot of the sessions covered a lot of real uh a lot of solutions that tackle real business problems. M uh so for me again that was very near and dear to my heart.
So overall a good a good balance of making things better and a good balance
01:57
of also bringing in new things to the platform. So yeah that's a really interesting point about everyone talks about the success of AI the bedrock of your data and the quality of your data and yet there's rarely any tools to kind of like discuss that or like the you know when I hear
02:12
those conversations it doesn't then transition into and here's how you can make your data better. It's like, h, okay, we'll just figure that out before you start.
Like, it's just something so trivial to do. Yeah.
Yeah. You You know, it it's you know, vendors generally speaking love to talk about products they they're
02:30
building, right? Uh-huh.
That's great. It's nice.
It's trendy. But sometimes they fail at seeing how that technology connects to the real business problems and the desired business outcomes.
that people
02:49
I'm a tech but more than that I'm a business for me if the technology is not connected to addressing a real business need and then I I don't care about it I don't spend any time looking into it cuz hey it's just theory I like practical I like
03:04
real I like let's go out and fix some problems with this yeah so so when you say you're a tech person should be known you're an MVP so I want to hear about yourself I want to hear from both of you about the MV VP program. Yeah.
So, a little bit about about myself. I am the VP of data and AI
03:21
architecture at a consulting company called Seb. I'm also part of the data bicks MVP program.
I'm part of the data bricks product advisory board and also an adviser to Sigma which is just won part part BI partner of the year award with uh with uh with data bricks. So, uh
03:37
a little bit about myself again I I'm originally from Honduras. So it's uh for me to be part of this conference has been a very unique blessing to me and uh really I'm just taking it all in but at the end of the day if you're one thing you want to take away from me is I like
03:54
to talk tech I like to talk about performance I like to talk about all these things but ultimately technology needs to solve business problems and I know I've already said that like three times but that's just how important it is for me it's less about AI less about aentic less about even Lake Flow or
04:09
Lakeflow or Genie which great technologies, but hey, okay, that's great. How do we solve things with those technologies?
And is I think it's fair to say that you've been fairly candid with your feedback about things like Delta Live Tables. I know you've given a lot of feedback to
04:25
our product teams that has been something that has evolved over time. So, do you want to talk about kind of maybe where it was, where it is now, you know, the changes that you've wanted to see for a while?
Yeah. And and I'll try to filter my mouth ever so slightly.
uh for the last 3 years basically since Delta live
04:42
tables launch I've been actually a fairly outspoken critic about them there were a lot of different flaws the development experience wasn't that great wasn't interconnected to the unity catalog and all these great things that data bricks was releasing quality checks were hit or miss and over the last eight
04:59
months I've been working uh providing a lot of feedback to the to the team behind Delta life tables they've been working with a lot of other customers, not just my voice, but a lot of other customers as to what they should fix. And I made a lot of my concerns public many times.
So you can
05:16
look back and you know don't don't just take me just saying something new. It's real real issues that now the the team Bilal and a lot of the producting behind uh LakeFlow have turned the best of Delta Life tables brought it into a better
05:31
development experience. So now you have LakeFlow which incorporates LakeFlow connect and uh Lake uh Spark declarative pipelines name changed just today so totally trippy but and then I think LakeFlow
05:47
jobs so till data bricks decides to name it something different so but long story short they're bundling out of the box a great pipeline development experience plus they announced lakeflow designer today which uh I saw some of the early mock
06:02
ops of it. So, it was great seeing it from where those early mockups were to where it is at today.
Uh, and one of the things that I love to talk about is data bricks is a fantastic data engineering and ML platform.
06:18
How do we turn it from being great for just hardcore techies to those analysts, those business users? And I think between I think data bricks one is one of the things that was announced today great makes it a lot more practable for
06:34
real business users. Um and with with lakeflow designer I think you're opening up the possibilities for not just a hardcore technical folks to thrive in but really just anyone.
So and so business users that means a lot of things to different people with different definitions. Who do you
06:51
imagine is the kind of person who would as a business user would start using something like data bricks one or something like the the uh lickflow designer? I mean I mean so I would say to two different kinds of people.
So in terms of the that's a great question by the
07:06
way. Uh I think so normally you think about data engineering as being to the left and you think of the analyst as being to the right.
the closer to the right that you are
07:23
that's what I would consider the business users again within the context of non- tech right but it's the further to the right that you can get the better it is for those business users uh so I think that you know with designer it makes building
07:39
things easier with u with the data bricks one it makes even further right it makes those users the the ones that are just consumers. Uhhuh.
It makes it a lot easier for them to actually uh take advantage of what
07:55
everyone is building because they don't have to be developers. Mhm.
They don't they don't even have to know SQL. They can have they can access from my understanding they can access data bricks one and really go there as their hub for actually make use of all the business tools that they have without
08:10
really getting into data bricks which I think a lot of people didn't want to have to get into data bricks you know especially those that are not coders. So yeah, absolutely.
Yeah. So people love your practitioner.
You mentioned you're very hands-on. So appreciate that.
I want to hear about like the product advisory number one.
08:27
And then you also been running benchmarks and just rolling up the sleeves. You're some some of the results, some of your thoughts on that.
Yeah. So so in a nutshell, like the I've been very fortunate that my job allows me my job at SEP allows me to really spend a lot of time researching
08:43
technology. Part of my job's description involves the fact that I won't always say the things that are favorable to data bricks.
My my job is to be as transparent because if I'm if I'm not saying those things, your customers are going to be saying it with their wallets.
08:59
Um so sorry I lost track of what the other part of the question was. Yeah.
Product advisory like that. So, so with so with that is um I I give give a lot of feedback and the team at data bricks listens to that feedback.
Sometimes it's in private
09:16
depending on the issue. Uh sometimes it's in private, sometimes it is based on the public uh comments that I make.
They'll reach out to me and ask me some questions and they might share hey we're fixing this sometime or we haven't thought about it or this is the why we can't fix something. Mhm.
09:31
Uh so you know we mean we meet once a year uh whenever here and throughout the year we have different sessions. So uh one key thing that you should probably know about me is that I did not start off as uh working for a consulting company.
I've been working with data bricks as a customer for the last like
09:46
threeish years of my life. Uhhuh.
And I just started working for a consulting firm two months ago. So I've been so I so I so whenever I'm saying all these things I I really feel it for the customers that perhaps I'm not dealing with some of this issues on a dayto-day.
Um I I know other people are
10:02
and that's why I'm very passionate about talking about them. Uh you also asked about my performance benchmarks or per I like to call them performance comparisons right because I think that uh I like to keep things practical.
So for those of you that don't know me I like to write a lot about different things. Sometimes I compare data bricks
10:18
against different platforms. Sometimes I com I compare uh different versions of different products from other platforms against themselves.
Uh my personal philosophy is that whatever testing I do uh I keep it real. If the tests don't
10:33
look good for data bricks on something I keep the I keep the queries that I ran. If the tests look great for data data bricks that's very good for data bricks right?
But uh might you know every time that you run a comparison there's always going to be a mistake that you make.
10:50
Sometimes you the data set you built from scratch wasn't as as good as you as you considered. So my personal commitment is that whenever I'm publishing this comparisons is keep it public, keep it reproducible for people uh so that that you don't just take my
11:06
word for it. You can run the test yourself.
And for me that's very important being transparent because again at the end of the day if something doesn't look good for data bricks guess what that's going to push data bricks to be better. If something doesn't look you know it's good for from in another platform or or something is good for
11:23
data bricks then that's awesome and then I know data bricks is going to push even harder still to still keep that edge because it's like hey we have an edge here why should we give it give it up. Oh yeah, it's a community that makes us all better.
Feedback from the uh customers, from partners, from MVPs, all keeps
11:40
iterating and making it all better. Part of the whole community.
Yeah. Yeah.
And in the latest one that I ran, it was very good. Honestly, uh some of the results for data bricks came turned out a little bit more favorably than I expected.
Uh on I' I've always considered data bricks to be better in the larger scales of data sets and other
11:56
vendors to potentially be better in the lower size data lower like in the in the millions instead of like the billions. Uh this the results that I ran turn out very very good for data bricks much better than I expected.
So yeah. M so you there are anytime anyone runs a
12:12
benchmark there's always like oh but you didn't realize the blah blah blah networking was wrong and you did the blah blah blah you know are there any one particular things that uh someone pointed out to you that they thought that you overlooked that you're like wow okay that is super niche uh do I need to control for that
12:27
well I think there's there's there's so there's always there's always one thing right sometimes there's valid sometimes there's not valid like I had ran one like maybe a monthish ago or so that there are some feedback that I got. There was one piece of feedback that
12:42
while it didn't necessarily impact the nature of the results and really the outcomes. Uh it it was around the data not looking as natural as it could be.
It was too evenly distributed almost almost perfectly even evenly distributed in different layers like the
12:59
same quantity for a product across almost consistently across every single month. And there's there's there's a real flaw.
I was still comparing apples to apples, which is important because if I'm okay if my flaws are still comparing apples
13:16
to apples. Yes.
That I'm okay if there's flaws. But then for me that I iterate from that.
I'm like, okay, I knew that was a problem. I didn't spend the before coming to to data bricks summit I spent about two weeks working on a newer version of my test that has a lot more controls as to hey this
13:31
is how you distribute the data more adjustable because for me transparency is very important so and and there's always going to be mistakes but at the at the end of the day uh if I can compare apples to apples every single time I'm happy with with the testing. Well, I think that also brings up an interesting topic about the
13:47
limitations of benchmarks. So, don't get me wrong, you know, it's a great place to get started to be like, "All right, who's really on the cutting edge at the moment, but ultimately unless it's solving like that particular data problem that you have back at the office, like the benchmarks are theoretical, right?" So, if someone was
14:03
to do something themselves, um, and they're like, "Hey, I really like what Jose has done. I want to do something similar." What would you recommend to for someone who as someone who has been running their own benchmarks?
How would you recommend someone do that with their own data back at the office? Well, I would say I would say this.
Take
14:19
the comparisons that I do or the comparisons that anyone does always with a loving grain of salt. Okay?
It's one more data point. Your results might might defair on that on that that same data set.
If you test almost immediately
14:35
the next day, it's probably going to be the same result. But with your own data though, test things yourself.
Don't just take my word for it. How do you go about it?
Honestly, that that's a very tough question because I know at the end of the day, you know, always reach out to the vendors you're considering, talk to
14:51
them, do do proof of concepts, but don't just take what they're doing for you, dig through what they're doing. Are they comparing things fairly uh and honestly, that that sometimes is the hard thing.
And at the end of the day, do your research. For me, so I started using data bras about 3 years
15:08
ago, uh 3 to four years ago or so. And one of the things that helped me succeed is that I spent about 16 hours learning data bricks.
And I was not a cloud ar well I was barely a cloud architect at that point. I had never really done data engineering besid besides Excel because
15:23
you know that you love Excel you make Excel puns everywhere. Throw that out.
If you follow by the way on LinkedIn especially if you want more Excel jokes have a follow. I mean look look at the logo.
I mean it's like it's an Excel spreadsheet. But long story short,
15:39
what helped me set me up for success for understanding whether data bricks or platform B and C were the right ones for me was I spent a lot of time learning about data bricks. spent 16 hours.
Yeah. Uh 16 hours worth of course content and that helped me up to understand, hey,
15:55
how do you best manage costs? How do you best build pipelines?
All this do the research. It pays off and then ultimately pick the platform that you believe has all the capabilities that you need, not just because someone is selling them to to you, but because you actually can verify that.
So that's
16:11
that's my honest recommendation. So yeah, thank you so much for being with us here today, Jose.
I'm sure we could have talked so much more um about all the different things that customers can do. Um but yeah, that's all we have time for.
So, thank you so much for joining us. And uh by and if by by the way if if
16:27
you're on LinkedIn uh if you want to follow me there on uh through I think it's going to be displaying on the screen. Love to connect with you.
So, absolutely. Okay then.
Sorry for interrupting you. No, that was great.
Thanks for having us. And by the way, to paint the picture like we are closing out the whole conference last year and this year
16:43
they're like breaking down the expo hall. So appreciate everyone uh sticking by.
Yeah. Okay then.
And so now to a video that we did earlier with Jimmy. Hello.
Uh good afternoon. My name is Jimmy Oi.
I am the global head of sales
17:00
development here at Conference. I joined Confluent about 3 months ago whereby I'm helping drive the way our SDRs help our business you know grow uh in different territories across APAC air and also um
17:16
air territory. Uh I came from data bricks whereby I was a global account executive and the journey here is extremely exciting because I get to work with the two best companies on the planet.
Confluent which helps us in terms of data in motion and data bricks
17:33
which helps us with the analytics aspects of how our customers can start to run their AI model. So it's an exciting time to actually have AI in real time leveraging Confluence as a platform.
17:49
Okay. Okay, and we're here at Summit Live at pretty much almost the close of the conference and like Ari said earlier, you can see around us how people are literally starting to strip things and take them down now.
Uh, so that's almost it for this year, but as we think about next year, we went round
18:05
earlier today and we asked a bunch of people why they'd want to come back next year. 100% come.
Uh, if you're thinking about it, you're on the fence, come. You're going to learn a ton.
There's great people and there's so many great companies and so many activities.
18:21
Something it's definitely an experience that you'll remember and you'd want to come back to. Definitely definitely come to the the event.
You're going to get to meet a lot of networking um other people doing the same thing as you and connecting with other data bricks users that have the same issues as me uh and discussing solutions to those
18:38
problems. So many wonderful seminars.
There's a great a lot of great swag and uh expo booths. Um and you might not know what you're looking for, but you're probably going to find it.
Oh, you should definitely come. There's so many different um industries here.
Like I'm in hospitality, which is definitely like not very common here, and I've learned
18:54
so much from all the different industries. There's really something for everyone.
Okay, and welcome back for the final time at Summit Live. We've had an absolutely fantastic week here, Ari.
It's been amazing. You are the
19:10
mastermind behind this, so I want to say thank you to you. Thank you.
Thank Thank you. What have been your highlights this week?
Well, being with amazing co-hosts. We have Carly, uh, we have you, Jason, Allison, who is in the breakout session.
So, uh, all of that is just a big honor,
19:26
but then also to meet with customers first and foremost. Thank you.
Virtual summits, hearing about all of the watch parties. So, it's just really the the whole community, partners, influencers, bricksters, product managers, founders, uh, influencers, all of that.
So,
19:42
it's been an absolutely jam-packed year. And with that in mind, we're going to say goodbye and we're going to see you all next year.
Have a great year. Bye.
Bye. [Music]
19:57
Bye. See you next year.
Bye. Thank you.
See you next year. See you next year.
See you next year. Here's my ride.
20:17
[Applause]