Yoni Eini - Being Customer Driven, & Eliminating Hierarchy

(Intro Narrator at 00:00:00) Hello, my friends. Today, Joel is talking to Yoni, the CTO at Upsolver, and they discuss how being customer-driven has been his key to career success, how Upsolver's data lake platform is an end-to-end data solution, and why Yoni prefers a company culture that doesn't focus on hierarchy. All of this right here, right now, on the Modern CTO podcast.

(Yoni at 00:00:26) Here we go.

(Joel Beasley at 00:00:27) This is the Modern CTO podcast. I'm curious. Can you share a little bit about your background?

(Yoni at 00:00:41) Yeah, absolutely. I'm Canadian originally. I was born in Canada, moved to Israel when I was 10, grew up in Israel, did the army. Started programming early, when I was 11 or 13 or something like that. Mostly focused on math, though, really. I did my degree in math. In the army, I was doing some programming and then eventually switched to more algorithms. I was a DBA. I was a data scientist for a while. So a lot of stuff around data, analyzing data, but at large scale. It's kind of what I've been doing. I moved to Finland about four years ago, so now I'm in Finland. And I started Upsolver, what is it now, seven years ago? Funny. Basically kind of scratching an itch I had in the army, where we were dealing with large amounts of data and the process around dealing with data wasn't super mature, even though you have a very, very broad knowledge base. So we kind of wanted to try to solve the hardest problems with data. And so actually, we pivoted in the middle. So in the beginning, we were actually doing advertising just because advertising has the largest volumes of data. You buy data by weight there, relative to other industries. And then, but very quickly, we realized that kind of our passion is really the infrastructure part. So we pivoted to being an infrastructure for big data, which is where we are today. A bit of background about me.

(Joel Beasley at 00:02:16) Oh, that's pretty cool. So you've been doing this for seven years now?

(Yoni at 00:02:19) Yeah, yeah, under the name of Upsolver for seven years. The current product, I would say about five years.

(Joel Beasley at 00:02:26) Tell me a little bit more. I'm actually interested in this, about the moment you realized that you needed to transition.

(Yoni at 00:02:33) So when we started Upsolver, when we got started, really, we had the idea of let's build data infrastructure. And we felt like there's no way we can actually get that funded. We can't pitch that reasonably because it's a super crowded space and it's very difficult to differentiate yourself. So we kind of went with, okay, what's easier? And this was 2014. So we went in the direction of advertising, which was pretty hot then, and deep learning, which was not yet really known, actually. So we were a bit ahead of the curve on deep learning, but kind of went to large data volume spaces, but something that we thought we could succeed in as a company. But it was always something that, in the end, it wasn't our passion. It wasn't the thing that we actually wanted to do. So I think the pivot—I mean, pivoting is always challenging. It's always painful. You have to fire a lot of people. That's really not a fun thing. But I think that while we were doing the pivot, and it was for all the right reasons—we weren't actually succeeding. We weren't making money in advertising. We weren't very good at that. But our data infrastructure was really good. So I think that the pivot was kind of easy in the sense that it was kind of dictated by just our numbers and our passions. So I think in that sense, it was really an upgrade. Doesn't make it easy and doesn't make it fun, though.

(Joel Beasley at 00:03:55) And how do you explain to people when, you know, you meet them and they're interested in what Upsolver does? What's your answer to that?

(Yoni at 00:04:03) I mean, of course, it's a different answer if it's someone in the industry or not. But in the industry, we're a data lake platform and specialize in just huge volumes of data and streaming data. So things that would normally be expensive or difficult or challenging to do—long development cycles—are going to be very easy and they're not going to get in your way as you build out your data infrastructure and the data that you want to use for whatever business use cases you actually have.

(Joel Beasley at 00:04:35) And then what if they're outside the industry? What are you telling them?

(Yoni at 00:04:39) So if they're outside the industry, I would say that we're a database. Well, I mean, outside the industry, in the end, people don't understand really what big data is or what streaming data is. So it's a bit more challenging. In the end, it is a specialized tool. I mean, database is already a word that many people don't even know. But, yeah, I mean, someone who is technically literate, I would still say that it's a big data platform.

(Joel Beasley at 00:05:05) Help me understand streaming data, because a couple things come to mind. You know, there's streaming from the browser, right? There's streaming video. What's the context for streaming data?

(Yoni at 00:05:18) Yeah, you're right. It's an overloaded word that also has many meanings. So when I say streaming data, and even within our narrow definition, there are a bunch of different interpretations to it. But it would be data that arrives at high velocity with a timestamp attached to it. So if I have data—okay, I have the Large Hadron Collider, and it created six petabytes of data during an experiment, but the experiment doesn't have a time aspect to it. It's just a giant chunk of data that now you're going to process as a single thing. I wouldn't consider that streaming data. It's large volume. There aren't many things that create large volume of data that isn't streaming, but the Large Hadron Collider would be one of them. But then if you have bank transactions or you have users visiting a website or you have anything that's basically—you could call each unit of data an event. An event always happens at a moment in time. I would generally consider that to be streaming data. Now, of course, you can talk about what's—what does high velocity mean? Does it mean the data is coming in all the time, like every millisecond? Or is it enough that it comes in once a day? Is that also considered streaming data? So these are the more fuzzy boundaries. But I would say that the baseline definition is data that is representing events that happened in the world over time.

(Joel Beasley at 00:06:45) Plus, it sounds cooler at the conferences. If you go to the Big Data Conference and you're talking about streaming data, that sounds way better than five years ago.

(Yoni at 00:06:53) Streaming, real-time, and big data. Yeah, yeah.

(Joel Beasley at 00:06:56) That's what we want, right?

(Yoni at 00:06:57) Yeah.

(Joel Beasley at 00:06:58) Do you get into—what do you do with the data? Do you clean it up? Do you store it? What do you do with the data?

(Yoni at 00:07:05) So when you say, okay, a data lake platform, that's—I mean, it's really kind of an end-to-end thing. So the first thing is that you need to store the data in the data lake. So you get data. It's coming in from somewhere. You know, it could be coming from, again, depending on your vertical, it could be browser events. It could be things that happen in a game. It could be advertisements. It could be bank transactions, you name it. But it's coming in and getting stored somewhere. Generally, that somewhere is going to be one of two places. Either a streaming data system like Kafka or, let's say, cloud alternatives like Kinesis or things like that, or it's going to be landing—it's going to be landing in a data lake. And I'm kind of taking a step back here and saying, I'm talking about the cloud, the cloud-native environment, so people that are streaming data in the cloud. If you're on-prem, things are going to be a bit different. But in the cloud, generally, it's either going to go to a streaming system or into S3 or into blob storage or something like that, into a data lake file. And then from there, we're going to pick it up, organize it. So we're going to kind of give you a curated data lake with your raw data, but then also give you all the tools you need to get it wherever you actually want it. Because in the streaming system or in the data lake, it's not actually providing any data services. You have the data, you're paying for it, but you can't ask it, am I doing well? Or ask it, how can I do better in my business? And generally, the people who are going to be asking it questions are using tools like Tableau that want to read from a database. So you really need to take that raw data, which is messy. It's often going to be JSONs that are nested. It's going to be disaggregated or needs to be looked up or enriched by other types of data. And only then can it go into the place where you're actually going to be consuming it. So Upsolver takes care of that entire value chain, starting from the streaming system, going into the lake, and then taking it all the way to the target database after whatever transformations, enrichments, joins, filters, whatever you want to do with it, basically.

(Joel Beasley at 00:09:09) Oh, that's pretty neat. What's with the name? Why did you call it Upsolver?

(Yoni at 00:09:13) Funny enough, it's a legacy name. I had another startup called Solver before that. And I wasn't—I didn't have enough cycles to think of a name. But, yeah, that's at least my interpretation. My partner, Ori, I think he thinks the name comes from a different place. So could be that one of us is misremembering something that happened seven years ago, but it stuck at least.

(Joel Beasley at 00:09:36) It sounds better than Solver, version two. Upsolver sounds better.

(Yoni at 00:09:40) Exactly. Yeah.

(Joel Beasley at 00:09:41) I like it. I like it. So when you're talking with customers, what is something that they don't quite understand at first when they're exploring your product?

(Yoni at 00:09:50) I think that the biggest, let's say, leap of knowledge comes from the differentiation between actual streaming data and data that's just sitting somewhere. So I think that most tools today generally are going to deal with a finite set—so they're going to be looking at all the data that I have without any kind of defined time boundary and without giving a lot of thought to when it arrived and how does that affect what I'm going to be doing with it. And I think that as people use our system, they kind of realize that the time element has a lot of information in it, and it's very important to the things that we're doing, because these events actually did happen at a certain point in time. And once they make that transition to thinking about their data as a stream—I mean, it is a stream. It was originally a stream. But then often their tools are going to force them to think about it as just a file or as a chunk. Then once you reverse that and you say, well, actually, treat it as a stream, suddenly things become a lot easier.

(Joel Beasley at 00:10:56) That's interesting, because you're exactly right. As I'm configuring the experiment or changing the variables in the experiments I'm running, those are all denoted by time, right? So if you're looking at all the data all at once together, you can't see the variations in the experiments too easily.

(Yoni at 00:11:13) Yeah, exactly. And a lot of things that are—I mean, if you're talking about data science, one of the biggest challenges when you have a dataset is that you want to label each of your events. So you want to say, a user arrived at my website. Did they buy or not? And so I might—I know that they bought because, you know, a day later, they went and bought something. So it's very easy to associate that. But what's difficult is to prevent a situation where I'm accidentally pulling data from the future for my predictions. And then if that happens accidentally, all my predictions are worthless because I won't, in real time, actually have the data from the future. So once you've chunked it all into one giant set, you lose that concept of data that's allowed and data that's not allowed, because you're not actually going to have access to it. It wasn't created yet. It didn't happen yet in the real world.

(Joel Beasley at 00:12:05) This is some pretty complicated stuff.

(Yoni at 00:12:07) Yeah, it's not simple. Yeah. I wish it was simpler. I wish it was simpler.

(Joel Beasley at 00:12:14) Well, thank you for doing it. I don't want to be doing this. So thank you for building the tools that we need, my friend. I really appreciate it. When your customers come to you, what is the feel? Do they usually know exactly what they want and they know that you're the solution? Or are they kind of wandering through this big data field trying to get some help?

(Yoni at 00:12:36) I think that, you know, we're not big enough for people to say that, you know, Upsolver is definitely the solution I want. I think that that's something that's going to happen as we grow and as there's more broader knowledge of what Upsolver is and what it's capable of. I do think that there's some educating involved. So usually, users are coming and they know they have this exact pain. They can really pinpoint—if I ask them, are you having trouble getting your data into the data lake in a queryable way? They're like, yes. Do you have to spend three months of big data engineers in order to modify a column to your output? Yes. The answer is yes. Is that a good thing? Definitely not. So they definitely resonate with the pains that we're solving, that we're helping—just avoiding, basically. But then they wouldn't necessarily know to say, I mean, the reason I'm having this pain is that I have to reason about data in a certain way today. I think that's something that's still—that requires kind of educating or just experience with the platform to kind of realize fully.

(Joel Beasley at 00:13:39) I love it. And what's the website? Can you spell it for me?

(Yoni at 00:13:43) Upsolver, upsolver.com.

(Joel Beasley at 00:13:49) Perfect. You just gave one of the best explanations of why people would need your software. So usually I do that at the end, but I'm like, you just said yes and no to different questions, and people are listening. And some of them are feeling that and they don't need to continue listening to the whole podcast. They just want to go to the website and talk to you guys.

(Yoni at 00:14:07) I hope it sounds concise on the website.

(Joel Beasley at 00:14:12) I'm curious. I wanted to know—we're in our, you know, our meeting preparing for this and we talk about data lakes, data warehouses. I wasn't sure what the difference was. Can you help me understand that?

(Yoni at 00:14:23) So I think that, you know, the big difference—and, again, all of these terms, this is in the end, this is my opinion. So all of these terms are super overloaded, and everyone says they're everything. It's not like—now there are also data lakehouses. So there's data lakehouse, data warehouse, data lake. That being said, my interpretation of this space is that you have traditional databases. That's just a database. So generally, it provides services of being able to query your data, has fast response times, but it costs a lot. It's super expensive per—let's say, a thousand dollars per terabyte. That's the main disadvantage. They're expensive, and, possibly, they don't scale that well. So when you get to billions of records, they're kind of choking. Some databases are going to be bigger than that, but it's a challenge, let's say, to have a database that size. And then the logical evolution of that is a data warehouse. So, basically, I'm saying, let's take the database. Let's get rid of all the stuff that doesn't scale well. So we get rid of the indexes and we get rid of the row-based storage.

(Yoni at 00:15:30) And so these are kind of streamlined databases, you can say. And the benefit is that instead of a thousand dollars per terabyte, they're gonna cost $100 per terabyte. So there's like a factor of 10 improvement. But I'm paying for that with slower query times. So it's good for analytics use cases, but not so much for transactional use cases.

(Yoni at 00:15:56) Well, they don't even have transactions generally. So they're not at all suited for transactional use cases. And then the data warehouse is still essentially a database in the sense that it still contains the data. It's still holding the data in a proprietary format in its own system. And then the data lake is kind of like a data warehouse—or let's say a good data lake, something that's not a data swamp.

(Yoni at 00:16:22) It's kind of like a data warehouse, except that the data is just vanilla files in a folder, and you have an external query engine that just knows how to work. So it's basically decoupling the compute and the storage. The storage is just sitting on cloud storage, on S3 or on blob storage, whatever it is. The compute is just servers that I'm gonna be spinning up or spinning down whenever I want to, but they're not connected. So I can say, well, now I want a lot more query. Now I don't need any queries. Let's just shut it all down. I'm just paying for the storage. And that's like maybe another order of magnitude. So where I was paying a thousand dollars per terabyte of uncompressed data in a database, and now it's $100 per terabyte of uncompressed data in a data warehouse, in a data lake, maybe I'm paying $5 per uncompressed terabyte.

(Yoni at 00:17:10) So it's really like a very big difference. I mean, nobody's doing it to save money, but you can hold a lot more data then. So you have much bigger, just higher data volumes, which give you much more accurate insights.

(Joel Beasley at 00:17:23) More data. That's what everybody wants. We want more data.

(Yoni at 00:17:26) Exactly.

(Joel Beasley at 00:17:28) I'm curious. I wanna switch the topic of conversation a little bit to talk about customer feedback because I was talking with this great person, Glenn Nethercutt, which I think is one of the coolest last names ever. And so he sounds mystical, but he's at this company called Genesis, and they mainly do software to modernize call centers.

(Yoni at 00:17:49) Right?

(Joel Beasley at 00:17:49) And I thought this was interesting because when we talk about processing user feedback, sometimes the most annoyed moments of my life are interacting with call centers, right? So I was like, they must have some good insight on that. And so we were talking about implementing feedback from their customers and how they do it. And I was curious—now I'm asking other people, like, how do you receive feedback from your customers, and then how do you implement that into your product?

(Yoni at 00:18:16) So we're very much a customer-driven company. Our customers are always kind of pushing the envelope of what you can do, and we very much want to, let's say, predict and respond to what their needs are. Because that's a bellwether also for what everyone else needs in a way. It's much easier to develop the right stuff when people are actively asking you for things that are related to it rather than just develop based on some kind of idea that you had that might be accurate or not. So we really try to keep the customer engaged as much as possible, like, between support channels and Slack channels and community engagement and just sales calls and whatnot. But we really try to keep solution architects and developers in the loop on a day-to-day basis with customers.

(Yoni at 00:19:05) So if a customer has a problem or something that they can't solve, at the very least, we're gonna want to understand why is this the case. Like, how can we do it differently? So, generally, if a customer has a pain that's core, that's something that, like, it makes sense to add to our product, I'm happy when we can get it implemented and delivered in days rather than, you know, months or years that I feel like would be more traditional for a platform like ours.

(Joel Beasley at 00:19:34) So are you talking with like the leaders within your organization about what they're experiencing on their calls? Are you spending time directly with customers? How do you stay involved?

(Yoni at 00:19:43) I mean, all of the above. I'm often speaking to customers. I would say every—I almost—there isn't a day where I do not speak to a customer. So I think that's super important, and it's the same for all of our management. So like, there isn't anyone on our management team that isn't, like, on a day-to-day level interacting with our customers.

(Yoni at 00:20:03) And, you know, it goes in both directions, right? So we're gonna be interacting with them and providing support and accepting their feedback and trying to make the product better. And then also, we're gonna get case studies, and we're gonna do joint meetups, and all sorts of stuff like that. So we really believe in the kind of joint success with our customers as well.

(Joel Beasley at 00:20:22) Yeah. Some of my favorite companies, I've noticed, it's just deeply baked into their culture. There's no tool. There's no specific thing. It's just they really care, and they've surrounded themselves with great people who also really care.

(Yoni at 00:20:35) Yeah. Exactly. Like, from my perspective, you know, I mean, time is a limiting factor. So, of course, you can't do everything. But, you know, someone just recently was categorizing companies. Like, some have community support on their website, some have Slack channels, some have Discord. I'm like, we could have Discord. Like, but maybe we don't necessarily have the bandwidth to have an actual dedicated Discord channel. But it's something I would really want to do. I think that would be super valuable that someone could just, you know, ask me a question or ask someone on our team a question.

(Yoni at 00:21:08) So maybe we're not there yet. But, definitely, our passion and our desire is to get there. We want to be more like, with more interactions with the customer in whatever platform we can rather than less. So that's where the real value comes from in the end.

(Joel Beasley at 00:21:23) Yeah. You're passionate about this.

(Yoni at 00:21:27) Yeah. No. Absolutely. It's like, why else am I building software for people? I mean, that's what the platform does. If people aren't using it, then it's just what? What's the value? I wonder where that comes from. You know, when in the army, I was there for nine years. So like, a lot of my professional background is from the Israeli army.

(Yoni at 00:21:52) You'd have a lot of stuff that's developed, not for any specific use case, kind of like someone thought it's a good idea. So, I mean, the repercussions of that are that often you're gonna get into a situation where you have these giant systems that are developed that don't have a clear customer or a clear use case and are often just forgotten. And then if you add to that that the soldiers, there's a high turnover. Like, people have their mandatory service, which is three years. Maybe some of them sign on for an extra three years.

(Yoni at 00:22:22) But generally, you have this up to six years between when someone joins the army and between when they're forgotten and never going to touch the software again. So often, you have these kind of dead systems that materialize. They're someone's baby, and they kind of grow to something that consumes a lot of resources and energy and passion. And then once they leave the army, they just disappear and kind of vanish into the ether. And then I think that, you know, it's on the other hand, when you develop a system that's actually in use and can actually, like, save people's lives, that feels super valuable.

(Yoni at 00:23:00) It's something that really gives you a kind of rush when you see that. And especially since it's so rare there that your systems are actually gonna see the light of day relative to the real world where, you know, companies that develop stuff that doesn't get used just disappear. So I think maybe that's where it comes from, that it's difficult to accomplish in the army, and that's why maybe it's kind of like a target to strive for and maybe something that then comes out of that. I wonder if many Israeli startups are like that.

(Joel Beasley at 00:23:33) I noticed they all have really passionate founders and a high level of discipline.

(Yoni at 00:23:37) High level of discipline. So yeah. No. You definitely have not been to the Israeli army. Alright. They appear to have a high level of discipline. They appear to have. Yes. That's more accurate.

(Joel Beasley at 00:23:50) I wanna talk more about leadership. What type of leadership lessons did you learn while serving your country?

(Yoni at 00:23:56) Good question. I would say that, you know, the army is kinda funny in the sense that you have a ton of super passionate young people who are really giving it their all, and you have like, kind of the card of, yes, you're doing it for your country. But on the other hand, they have to be there. They don't have any alternative. They don't have an opportunity to say, well, no. This is not what I feel like doing. So a lot of kind of leadership—I think it's actually quite that difference that in the army, it really takes leadership to activate your soldiers. Whereas in industry, it's like, leadership is also important, but that's more like, I would say, a C-level thing. And then management is very important to make sure that people are actually happy and—and there's not really a concept of management in the army. Because again, people have to be there.

(Yoni at 00:24:48) They don't have the alternative, which they can just, like, you know, get up and go. So management really isn't a forte or anything like it. It isn't something that people focus on there. So you have a few leaders that are gonna make people passionate about projects or passionate about ideas. And then you have a bunch of people that aren't, in the end, are either self-managing or they don't really succeed.

(Yoni at 00:25:10) And I think that maybe the thing that I would most learn is that you actually need management, unless you're hiring directly out of the army, and then you're gonna end up getting self-managed people. Aside from those, you definitely need management principles that—I don't know of a lot of organizations that manage with just self-manage. Maybe, I don't know, like, things under Elon Musk or something like that. But like, for the most part, like, you do actually need management. And that's not something that the army is very good at.

(Yoni at 00:25:39) So I'd say it's the anti-pattern that I learned there.

(Joel Beasley at 00:25:42) And what have you learned is some of the most important things? Like, you've got managers at your company. What do you teach them?

(Yoni at 00:25:49) I think that in the end, you know, it's very common that you're gonna reduce your team to like, kind of a bunch of numbers. I mean, you have to in the end to some extent. Because you have X developers and you need to do Y tasks. So you divide X by Y, and then you know, like, kind of how many developers are gonna go to each task. But it really doesn't work that way.

(Yoni at 00:26:14) It's not like—individuals are way too individual, and they have different competencies. They have different desires. Like, when you reduce it to just numbers, it kind of looks like it works, but there's a lot of complexity there that if you happen—it's very happenstance if you got it right. And I think that—and this isn't something I needed to teach any of our managers. Like, we're very lucky to have managers that just get this.

(Yoni at 00:26:40) But that, you know, people aren't just interchangeable and each individual has a lot to contribute in and of themselves. Yeah. I think from my perspective, being an Israeli company, this is actually—you're asking about positive leadership patterns that I learned in the army. One of them is that there is no—it's funny because you'd think it would be the other way around. But like, as opposed to the U.S. army or other armies that I know from from movies, in the Israeli army, there is no such real thing as hierarchy.

(Yoni at 00:27:15) So you have the hierarchy. You have your commander, and your commander tells you what to do, and you basically do what they say. But you can yell at them if you want to, and you call them by their first name. And even if you're talking about like, you know, the commander's commander and their commander and like, the commander of your unit and of the whole army, people are all on first-name basis. It's kind of odd.

(Yoni at 00:27:35) I mean, you know, our prime minister is like, he's called Bibi. That's his—it's a nickname. Like, that's how he goes by. So that's, I think, very much the army culture that, you know, anyone at any level can talk to anyone at any other level. The doors are all open.

(Yoni at 00:27:51) I think that's something that you see a lot in Israeli startups, and I think that creates huge value that someone can tell you, hey, you're an idiot. And nobody gets offended because, you know, everyone does it. So that just means that he cares. I think that's something that's hugely beneficial. I can't imagine how a company can succeed if they're not working that way.

(Joel Beasley at 00:28:12) Yes. I love companies with direct open culture. The easiest to for me to exist within and to understand because I can't do the eggshell dancing around thing. I just say what I feel. You know?

(Yoni at 00:28:28) Yeah. Exactly. Yeah. It's kind of refreshing to find in Israel—again, everyone's like that. It's just kind of like a army culture or a general culture thing.

(Yoni at 00:28:37) So it's—but outside of Israel, finding people who kind of subscribe to that, I think, is not rare. But, I mean, you have to look. And I think that it really—it's something that can be taught as well and is very valuable for cultures that don't have a lot of ego, I guess. Or for, let's say, for people who don't have a lot of ego and can accept that someone is calling them an idiot and not take that to actually mean that they're an idiot. It's just that they had a bad idea in someone else's opinion.

(Yoni at 00:29:05) Yeah.

(Joel Beasley at 00:29:06) Yeah. I find it a lot. Every time I see on the preps that I'm talking to someone that was in the IDF, I get excited because I'm like, oh, this is gonna be a great person. But I also find it in executives at modern companies, and I use that term very loosely. It doesn't necessarily mean startup.

(Joel Beasley at 00:29:23) It's how the company—the culture of the company. Some companies are more modern than others. They're adopting ideas faster. And this concept of open direct feedback is something that I love seeing spread throughout, you know, the ecosystem because I think it's the winning formula. You can get things done faster.

(Yoni at 00:29:42) Yeah. I totally agree with you. And yeah, it's like kind of—it's nice to see all of these companies that aren't—like you said, they're not small necessarily, but it feels like I don't know. Maybe it's like a Silicon Valley mentality or like, I don't know where it comes from exactly, but it feels very different from kind of the image of the kind of large corporation where, like, you don't really know even who's who's necessarily above you there. Yeah.

(Yoni at 00:30:07) I wonder if it's just a matter of scale. Like, in the end, all of these Silicon Valley companies are, even if they have large revenue and large—but they're actually quite small often. They don't have that many employees in the end. So that might also be a factor there.

(Joel Beasley at 00:30:24) What's the tech ecosystem like in Israel?

(Yoni at 00:30:27) It's, as you would define, modern, I would say. Like, in the end, it's a very small country. But so that means that on the one hand, you have everyone knows everyone else. So you have—you have a lot of startups there. But, generally, startups are gonna know each other.

(Yoni at 00:30:45) They're gonna know what the other ones are doing. So you have a lot of cross-pollination. People are gonna be each other's customers. They're gonna be mentoring each other. Like, there's always some—like, either you work with them, you were in the army with them, or you were in the army with someone that knows them.

(Yoni at 00:31:00) So it's a very kind of neighborhood-y feeling, I think, and that really affects kind of how everything looks. I think that's like the predominant feature. It really comes out of the army. You know, the army is rejecting thousands of people every year. Not rejecting—they're just, you know, they're getting out of the army and ready to like, they have all the job experience necessary to start the startup. So that's exactly what's happening every year.

(Joel Beasley at 00:31:30) We get a lot of questions about, you know, how to grow and develop yourself from, you know, the next generation of leaders, right? So we've got all sorts of people who listen to the podcast, but I get a lot of outreach saying, you know, what's the advice? What's the one piece? And I'm always thinking, well, there's a lot more than one piece. But I was curious if your engineers that—or let's say engineers that are listening and they're aspiring to take on more management responsibilities—what sort of insight do you have for them?

(Yoni at 00:32:03) I think that the most important thing, like the biggest differentiation between an engineer and a good engineer that's a manager, is that the good engineer that's the manager is looking at things a lot more holistically. So an engineer is generally going to be trying to solve a very specific problem. And like, you know, I do that too. It's really fun to try to take a puzzle and just, you know, solve that puzzle really well. And I think that, like, you know, in the end, the companies are built on that. That's what they need is people who know how to take a puzzle and really solve it.

(Yoni at 00:32:35) I think the main difference and the main kind of thing that—the mental leap that you need to do in order to go from being an engineer to either a manager or an architect or, not just an architect—like an architect or a manager. I would say these are like two parallel jobs, not like one below the other, is to kind of try to look one level above and ask, "Why am I solving this problem? Like, why is this problem important to someone who's going to end up having a better life because I actually solved this problem?" And maybe the answer yesterday was, "You know, I think these people, but I'm not sure." And then today it's, "Well, actually, no, it doesn't look like anybody cares." So the problem is still as compelling as a technical challenge, but it doesn't have any real—I mean, you know, calling it—I think a lot of engineers, if you call it business value, that's kind of like a swear word. Like the business value is not really kind of what they're—but I mean, maybe look at it from the customer's perspective. Like, you want your stuff to be used. You want people to enjoy it, to feel value from it.

(Yoni at 00:33:43) I mean, that's sure, it's business value in the sense that that's what, like, you know, in the end, that's what the business is going to get paid for. But also it's kind of like customer satisfaction or enjoyment or amazement. Like, all that comes from these little features that you do if they're valuable, if they end up affecting something. I think making that step up and saying, "Why am I doing this? Like, what is the goal that we're actually trying to achieve? Who's going to enjoy it?" I think that's a very important thing for an aspiring manager or architect or CTO to think about. Because that's what you think about all day when that's your job.

(Joel Beasley at 00:34:23) Yeah. We're always trying to figure out the better questions to ask and where can we resolve tension in the processes and—yep.

(Yoni at 00:34:30) Yeah. Yeah. And hopefully you have a bit of time left for coding. That's—

(Joel Beasley at 00:34:36) You know, I like that because what it alludes to is that you have to do the job at the next level before your title changes.

(Yoni at 00:34:44) Yeah. Yeah. At least—yeah. Because—bring value to it.

(Yoni at 00:34:48) Yeah, it's true. Like, in the end, the people who are going to be promoted to being managers are the ones who are already kind of showing that kind of forethought. Because, you know, I mean, the managers don't have that much imagination. They can't see what you're going to be when you're given that responsibility. So they have to kind of already see it in how you're doing it in order to kind of imagine, "Oh, okay, so then it's very natural that then I'll make them a manager because they're already thinking about things in this more holistic way." Yeah. You have to be the next step.

(Joel Beasley at 00:35:23) Yes. Yeah. You said it better. You have to be the next step. I like that. Because it's exactly the truth, you know? And when I'm—I'm trying to think right now about my team and it's never, "You will get promoted and then I hope," because that strategy doesn't work. It's always, "I see potential in you and I'm going to empower you farther."

(Yoni at 00:35:46) Yeah. Yeah. Absolutely. I mean, I think at least for our types of organizations, I think there are like the bigger organizations where it's a lot more rigid and a lot more, I think, difficult. And then possibly, if you're acting like a manager but you're not a manager, that's not a good thing.

(Joel Beasley at 00:36:03) Well, we have a thing for the show where like, you have to come with common sense. Because the people who are going to misinterpret it that bad, I just—I get to go out, you know, before the whole pandemic and go meet the people that listen to the show at different talks in different cities around the world. Everyone's super bright. So I feel like this is a good space to give advice and then people—but there's still going to be one person out there that messes it up.

(Yoni at 00:36:32) That's going to misconstrue. Yeah. Yeah. Yeah. Yeah. Yeah.

(Joel Beasley at 00:36:35) Oh, dude, this is great. This is great. Now I want to make sure that we touch back a little bit to Upsolver and its mission before we wrap up here. So can you just give me the brief overview? Actually, what I want to know specifically is, have you found that there's any industries that use you more? Like, do you excel in health care? Do you excel in marketing? Or is it completely agnostic to the industry?

(Yoni at 00:37:02) So it's not completely agnostic to the industry. And I would say that if you're looking at an industry, like a vertical or—so I would say definitely digital native businesses, like businesses that were born on the Internet, that were natively in the cloud, that, you know, what they do is process data that they got from somewhere and send it to somewhere. And that's actually what they do. So advertising companies, e-commerce companies, gaming companies, all of these I would consider like digital native. I mean, this is actually a word that's used in the industry, but like—yeah, it's, anyways, like any other word, people have all sorts of different definitions. But I would say that digital native businesses and ones that have more data than they can handle or more, let's say, more data needs than they can handle. So they're struggling to get more big data engineers. They're struggling to get more developers. They're trying to grow. They're trying to provide more data services and not necessarily succeeding as much as they want to. I think that's kind of our sweet spot. That's a place where you're like, "Yes, Upsolver exactly solves my problem."

(Yoni at 00:38:02) I mean, that's not to say that other people don't need Upsolver, but I think that that's like the slam dunk, is that I want to increase the velocity that I can provide data services. And I can't solve that by hiring 200 engineers because either they're too expensive or I just can't hire 200 engineers that quickly. So then you want a platform that really kind of makes the job easier.

(Joel Beasley at 00:38:35) What about trust? Do you have access? Can you see my data? Do I trust you with my data?

(Yoni at 00:38:41) So we don't take any kind of access to your data. So Upsolver is deployed as infrastructure as a service into your account. So you're running on AWS. You have your own VPCs, your own AWS account. All the data is stored in your S3 buckets. All the data stays within servers that are in your account. So Upsolver has no access unless you grant us access, unless you give an Upsolver support person an open port so they can look into that environment. They're just not going to be able to see anything. So the data completely resides in your account. I think that's, for most use cases, that's super important because, you know, who—yeah, who wants other people to see their data? It's like, not—yeah.

(Joel Beasley at 00:39:25) Yeah. I like it. And it helps with compliance. It helps on your side because now you're not storing it, so you don't have the responsibility of owning it, right?

(Yoni at 00:39:35) Yeah. Absolutely. And I think that most—you know, it used to be that way. It used to be that, you know, you paid Oracle, but it was installed on prem. And it's only with the migration to the cloud that suddenly vendors have started feeling that they can appropriate your data. You send it to someone else's account and they're kind of like, "Okay, it's yours, but I'm holding it for now." I mean, aside from Amazon, which are clearly holding everyone's data, but that's just the way the world is right now. But like, putting the three big cloud vendors aside, I mean, I'm not sure it's legitimate for a vendor to hold your data. Why would they kind of—like, it needs to be very compelling, and I can't really think of a good reason for that to happen.

(Joel Beasley at 00:40:18) If people want to learn more about Upsolver, they want to reach out, get a demo, explore it, how would they do that?

(Yoni at 00:40:24) Yeah. So if you want to learn more about Upsolver, you can go to our website, upsolver.com. You have tons of information there. We have white papers and use case studies and all these things. We have a community edition, so you can just go ahead and start using Upsolver. You don't have to pay us anything. We're not going to call you or anything like that. You can just go onto our website, sign up, and start playing around with it. We also do dev days periodically. So if you want to kind of have a walkthrough where someone walks you through a use case that's kind of been tailored to kind of get to know the system and get to know data in general. So we also have those. You can also find the links to those on the website.

(Joel Beasley at 00:41:04) I love it. I love it. Alright. So now we did it. We made a podcast. How'd you feel?

(Yoni at 00:41:10) Awesome. Woo hoo. That was nice.

(Joel Beasley at 00:41:29) Thank you so much for listening. And if you found this episode useful, please share it with a friend or colleague who you think would get value from it. And if you have topics that you'd like to hear discussed on the podcast, either add me on LinkedIn or send me an email, [email protected]. Every time I get an email or LinkedIn message, it absolutely makes my day and inspires me to keep going.