All right, welcome to another episode of Energy Bites. We got Brian Weaver here from Liberty Energy. He's the head of software development and data architecture. Yeah, thanks for joining us, Brian. Yeah, thanks for having me. And we got John Calfein here, the rad dad. Yeah, with the [snorts] uh torn-up backyard from the Yeah, we're we're dadding [snorts] today. Colin's is at the office with me. Jake's kids are here. His wife's here. No one has power. It's fun. Um But yeah. Energy's kind of important. Yeah, shocker, you know? My Yeah, my daughter was like, "It's not It's not cold in the house." I was like, "Yeah, that's because we don't have power." Yeah, all that stuff I've been telling you about how important energy is, this is why. Yeah, see it It's not coming from the wall right now. [laughter]
The magic isn't happening anymore. [laughter] But no, we're we're all good. We got power last back last night. I know people are probably not going to have power for potentially days to come, so You mean people with like the transmission lines that are just folded in half? Say hi. Okay. Now, speaking of She's She's running part of that. There she is. She ran away. [laughter] Yeah, no, those those were crazy. I've never seen transmission lines like just crumbled up like a piece of paper before. Yeah, downtown Houston looked like a war zone. You got windows blown out and It's insane. But uh Well, cool, but Brian's in Denver, so it's not his problem. Um No, we're pretty boring here when it comes to that stuff. Awesome. So, how are things going over at Liberty Frac? Doing well. It's been uh been a busy year. Um had quite a bit of growth over the last few years, buying some companies. Bought Schlumberger a few years ago. Um that was a pretty big integration, so everything's kind of nice and boring, which is been a little unusual in our company. We're finally at a a steady state for a little bit until we do something else. Although, we are moving to Australia. We're sending a fleet currently over to Australia right now, so. Oh, very cool. You have anything else international currently or No, mostly Canada. So, we're we're really more of a North American service company and this is kind of our our first dip into the international waters. Although, we are and I guess we have some stuff looking at in Central Europe. Okay. So, we kind of see how those go and if that expands. Nice. Yeah, I don't know if there'll be any Vaca Muerta down there and South America or anything, but Yeah, that's the other big play these days. I think I just saw some production stats coming out of Argentina that are pretty impressive right now, so. Yeah, I think it was Ted Cross who we had on the show that he released that stuff and yeah, definitely looking really competitive at least with the Bakken or Eagle Ford if not, you know, if not the Permian, so. Just Just the rock though. Like, I mean, it's so much bigger [laughter] than I feel like anything we have stateside. Yeah, the Eagle Ford has you know, been prolific, so was the Bakken, but the Vaca Muerta just looking at the thick Just the thickness, right? Like, there's Yeah. Sections that are thousands of feet thick. Like, that's crazy. Yeah, and then, and and then we, you know, we talked about this, too, but I mean, the advantage of being that I mean, not even fast follower, but even more of a laggard, but I mean they've let us take, you know, 10 plus years to 15 years of of learnings, and now they can go in and, again, and then they don't have all the the land, you know, complications like we have here. You know, so it's not like you have to worry about a unit well or, you know, this, you know, crossing over everything. It's not I mean, you want to drill drill three miles, drill three miles. If you want to stack them, you know, wine rack them, however you want to do it, uh you can do it, so. Yeah, I'll be interested to see how that plays out. It's uh a huge opportunity. Hopefully, the government can stabilize and Mhm. make it more attractive for a lot of companies down there. Yeah, yeah, especially I mean, I think we we always say that oil field services is kind of one of the limiters there, too. Like, you know, the economics work here cuz you can do it affordably. But you know, as soon as you start needing, you know, you guys or Halliburton or someone to mobilize or set up fleet down there, it becomes a lot more expensive just be what you're going to charge, you know, for that stuff. So. Yeah, I worked in Australia for a year and that was one of the challenges they had there was they don't have that you know entrepreneurial like when we have a problem in the United States a company pops up solves [laughter] that problem. Down there it's tougher like you just can't jump on that stuff and so and then the [clears throat] other issues in some of the international markets are logistics. We have a rail network that can move pipe and sand and whatever we need pretty fast. Uh keeps our cost low. Over there it's a terrible road to drive out there and it's some truck driver and you hope he makes it and it's it's much more challenging. And sometimes you have to hire the locals as well, right? So, I mean then you, you know the training issue makes them...I know we're getting saturated here. We got training issues as well. Just, you know, got green hats all over the place on frac fleets and on the drilling rigs, but, um... So, I mean, I think this might be a good foray, like, you know, jumping in on some of the tech side of it. You know, we kind of sometimes handle this kind of non-linearly; we'll kind of work backwards. But I mean, with you guys expanding, say, into Australia or even like Central Europe, I'm, I'm sure you've got, you know, telecom type stuff set up here where you're, you know streaming, data and I know you do, and all that going on but you know, how's that going to change, maybe from your side, from the, the data and data architecture and that side of it? You know how's that going to affect or how much are you know are you guys even going to focus on that side of it moving into those markets? Yeah, we do focus on it. The biggest yeah as you mentioned the biggest issue is connectivity. Starlink is getting there but it's not there everywhere in the world yet. Um even in Australia we only get like 8 hours of coverage where our fleets are going to be at. Although they're about to do a launch here in this summer that might help some of that coverage. So but there's no cell phone towers out that way. The nearest town from where we're going to be is like an hour and a half drive. Um so there is the remoteness does add a lot of complications. We'll still collect the data we'll still be able to process it. It just might not be as in real time as we've been doing in the past, but once we have that connection, everything else is in place. Really, the biggest issue is the connectivity. Everything else pretty much flows pretty seamlessly with the cloud infrastructure we run. Nice. And is that you know, so even like stateside or do you focus heavily on the on the satellite now that you have some of the lower bit stuff like Starlink and all that stuff or do you try to leverage cell phone where you can, cellular where you can or Yeah, it's a bit of a mix. Really, we whatever's the best signal we'll leverage. So, we do have these boxes that kind of switch between both Starlink and cell phone coverage. Um for example, here in Colorado in the DJ, we have ridiculously good 5G coverage in our field. Um West Texas definitely has spots, but it's it's a huge basin. It's not fully covered. So, um yeah, I think we kind of flip between the the two, but I mean, when I started in the industry uh 18 years ago, we never had cell phone coverage. We had a sat phone that cost a dollar a minute that we're told never to use unless you had to, kind of thing. Now, it's a complete opposite. Everyone's out there on their phones. We're we're streaming quite a bit of bandwidth out there, both personal and for work purposes. Um it's definitely like as far as connectivity, it's it's amazing how far we've come in 10 years. Yeah, I mean, right now I'm I'm operating off of 5G home internet and I'm getting 300 megabytes in my house, you know, cuz I I moved out of the country and the best hardwire they had was 50 megs. It's like that's not going to cut it. And yeah, Verizon 5G, there's a tower right here and I'm going to stream 300 megs from my house with just a cell modem essentially. It's it's crazy. Yeah, the rural areas really I mean, we've we tried to invest all that infrastructure in putting copper out to the rural areas, but satellites like Starlink and some of these others, the 5G stuff is basically going to replace it all. Yeah. Yeah, I mean, John and I were kind of on the pseudo frac side, you know, I I guess it was about 7 years ago and it was we barely used satellite only at a pinch, and it was slow, and it was expensive. And then again, again, we're talking 6-7 years ago. It wasn't Starlink, either. Yeah [laughter]. It was all the incumbents that, you know, would just murder you over their data policies and and plans, and it was terrible all around. Yeah. You'd have to set it up ahead of time. You couldn't have any kind of active switching, all that fun stuff. Yeah. I worked for a microseismic company uh Pinnacle quite a few years ago, and we had satellites, but how we did it is, you know, we're pulling geophysical data, so it's huge quantities of of time series data coming in. Um and we had a trigger processor on the the equipment, and then we only sent those, you know, what triggered events were sent back for processing. Um and we had we literally shipped the rest of it on a hard drive cuz it was just such large quantities. There's still folks out there who are still doing that cuz when you get into the, you know, terabyte, even you know, higher than that, Yeah. it's hard to do it over the internet. Yeah. Yeah. And do you really need that information, you know, in real time, you know? [laughter] Yeah. Yeah, most of those hard drives ended up in a closet somewhere anyways, locked away for it just in case reasons. Yeah. [clears throat] So, I guess I mean, again, I think probably makes some assumptions. John and I are pretty familiar with the frac space, but, you know, could you kind of elaborate on, like, the type of data that you guys are, you know, capturing and, and just kind of, like how, how it's being used, like as far as, like, Liberty goes? Cuz of it, I mean, there's obviously a lot of vendors, like, you know, even RDS, where we were at, still will do some, uh, frac, you know, data capture, and there's Cold Bore and Corva and all these different people. Um but I mean, I remember it was probably 2018-2019, and and you guys were kind of on the forefront of being a frac company that was actually capturing this data, and I think we're putting it in a Tableau dashboard and some of that stuff. So, uh could you kind of just give a high level about how how you guys are, you know, pulling data in and how it's being used, how it's being served out? Yeah. So, I mean, one of our advantages we're a pretty young company. So, we don't have a lot of data that like that old legacy stuff. Um so, we've been capturing it in databases for pretty much from the beginning or we were able to get it into a database pretty quickly. Not not a concept. Yeah, it's really nice cuz I you know, I worked for large service companies like Schlumberger and Halliburton and yeah, some of their stuff is in closets like I just said. Yeah. Never to be found again. But um Yeah, we really focused on, you know, there's kind of two main sets of data coming from the field operations. One is collected by our field engineers. So, they collect in different points. They're collecting volumes, averages. They're doing some calculations for the clients. Um all that stuff is captured and then sent back um within about 5 minutes of a stage ending um back into the our cloud infrastructure. Um that data we've been collecting since the beginning in different forms. Um I want to ask you about that. I want to ask you about that because when I was a frac engineer, that was a post job report, right? You know, in in one of the the three original, you know, Excel workbooks that everyone copied and took to every new frac startup [laughter] that existed after that. But uh How did How are you One of the, you know, that has all of the good data in it essentially, right? It's all the summaries, all the averages. Right, it's all the important data. But, you know, we would put it in Excel and then we generate a PDF and then ship it off to the client because that's what they wanted then. How did y'all um I mean, are y'all doing that in like a form-based workflow now or is it still very similar where it's, you know, it goes into some kind of standardized document and then you guys are able to parse it on the back end? No, so we're I think it was about 7 or 8 years ago when one of the engineers started exporting that data out in an XML format. So, basically we captured what was in Excel, put it into more of a structured format, semi-structured format, and then it was sent to a person in Houston. She kind of managed the processing, the intake. So, she was also the kind of the QC-er of that data, making sure it came in, you know, correctly and and but that was once we got it into an XML format, that really accelerated that databasing. Um to where we do a lot of kind of uh you know, decision-making off of that data. So, where a lot of executives, a lot of sales people, a lot of folks pay attention to that data, including our clients as well. So, that's all that post-job data is is very quickly accessible to us. Um recently we started moving to more of a JSON-type structure that we're pulling that data out with. Um and that's becoming a little more real-time. So, as the projects or the stage is going on, we're collecting that data periodically throughout the stage. And then once the stage is all wrapped up, uh we kind of get a trigger to let us know the data stage is done, and then that gets dropped into our database. So, yeah, we've we've been pretty fortunate. You know, we did start out that same way where it was sitting in an Excel document, and it was pretty challenging to get that data out, but we quickly pivoted from that. Um so, anyways, that's kind of our main data set, one of our more important ones. We're also getting into streaming of the frac data, so the 1-second data coming out of the the frac van from our equipment. Um that project is, you know, relatively new. We've we kicked it off about a year and a half ago, um and it's starting to gain traction now. Um so, that's from the field engineering side. That's, you know, kind of our main data set. We also collect telemetry data off of our equipment for maintenance reasons. Uh Caterpillar's been really helpful um on that side of things. We're able to capture uh you know, engine diagnostics, transmissions, uh power ends, things like that are all captured, sent back. And then now that we're getting into electric uh or electrified fracs, uh we're getting even more data points uh from that data set. So, all of that data gets uh dropped into an influx database that we use for a lot of our kind of uh diagnostics on our equipment and trying to do our preventative maintenance and uh that's probably one of our true big data sets uh that we capture from the field. Um and then we're a larger the What's the resolution on that? What's the Uh it's I think the frequency is about 1 hertz on most of it. Um we do process it because some of the stuff you don't need 1 second, you know, resolution on that, so we kind of do averages over time and that kind of makes the data a little more manageable for analytics. Um, but like I said, we're also a large enterprise, so we're collecting data from our ERP systems, from our HR systems, you know, sales systems, things like that. So, uh, that's been, uh, one of our big things: is trying to get that all kind of accessible to the company, uh, where it needs to be. Nice. So, all right, obviously there's a lot of ways I I would like to go with this as a as a data guy [laughter]. Um but you brought up you brought up InfluxDB, which uh I don't think we've had brought up on this uh on the podcast yet. And one of the big things we always try to do is like highlight some different technologies that people are using that people may not be aware of. So, could you kind of give a breakdown for people like what is InfluxDB and, you know, why'd you guys choose it and what are what are what are the advantages? So, Influx has a few advantages. It's a true time series database. That's what it's built for. And when you do queries out, you can kind of truncate that data, or bring it in ways that aren't in time series format that are actually usable; especially when you're talking about, like, if we have 1 second frequency over 6 months, it's not, you know, reasonable to bring that into a a BI tool. Uh so, Influx gives you tools and the query engine that allow you to pull it out in much more manageable ways. Uh it's also optimized for storage, so if the value doesn't change from the previous second, it just doesn't store it. It leaves it leaves it blank and it's assumed that it's that value before. So, if you have something that's sitting at, you know, 120 degrees for 3 days, you're only having that data point once. Um so, it's a very uh ingenious honest honestly ingenious system for time series data where it's optimized for both query engines and for the storage of that data. That would have been helpful John when we were messing around with D fits and you know, every engineer was telling us, "Oh, we need one second data. Like what if it changes? Like it's not changing. It hasn't changed the PSI in 5 hours, you know?" Yeah, but Yeah, especially D fits in these tight shells. That could take forever to walk away for a closure. Oh, no, that's [laughter] the that's the best part is they want they want Well, Bobby and I were building our dashboard for this because we had the ability to stream this data. And so you know, people don't need to see the one second fall off of a shale. And you literally can't put that many pixels on the screen. Like Yeah, that was the fun part, right? Like we were capturing all the raw data and then Bobby and I came up with a little, you know, filtering method. Ultimately, basically same thing that the database does. If it doesn't change more than 0.02 whatever the thousandth PSI, then don't add a new point until it does because there's only so many pixels on this stupid screen. And one second you know, one second data for a month, you can't even open an an Excel sheet, right? Like which is of course whatever you end user wants to do. Yeah. [laughter] And so Yeah, it's a People don't Those are all the things like people just don't know until they get into it and they're like, "Oh man, this is a lot." [laughter] Yeah, time series data is probably the most challenging data that we deal with. Just how do you manage that scale and make it reasonable for analytics? Right, that's the thing, right? Because ultimately you're you have to visualize it because that's how humans use data and you know, or that's the easiest way for them to make sense of it most of the time. Especially on like a frac job, right? Where it's like, "Okay, I need to know." And then it's not just Yeah, we we got lucky because for us it was just pressure and rate most of the time, right? It was just two data streams. It wasn't 50 different uh you know, frac channels and then wireline and pump down and then all the equipment information too. Like I can only imagine how how much fun that is for you guys. Yeah. But I am glad to hear y'all are pulling all the the, you know, equipment data too. Cuz as someone who was a frac engineer, like I thought it was I, I came from the manufacturing, well, my two internships before I got in oilfield were in manufacturing, and so I was hammered, you know, 5S, 6 Sigma, you know. My, I would literally stopwatch everything, you know, trying to shave seconds off of the process or minutes, and then I get in the oilfield and it's like, "Oh, we don't monitor." We're writing, you know, the last time the valves and seats were changed in chalk on the pump and that's it. And then each supervisor goes by and checks it when they come on location and that's our analytics. You're just like, oh, okay. There's lots of room here to uh Cuz everyone wants to like the shiny cool stuff, but it's like, you know what also sucks is being shut down cuz your pumps need maintenance and you didn't do the maintenance. Yeah, that's a good way to lose a contract. Yeah, real quick. Yeah. Well, and then the scale of those seats and valves gets pretty big when you're at our size and so if we could ex-extend it one more stage or you know, whatever without costing downtime then yeah, it's it's pretty valuable to us. So yeah, we try to capture everything we can on that side. Yeah, no, the sk-that's the thing. Like the scalability of frac is so crazy when you really start looking at like putting the numbers together. It's like, oh, 5% on one stage, not that you know, not that huge of a deal, but 5%on one stage times how many stages are you doing in your as an operator this year? Like that's a lot of money. That's millions of dollars more than likely. Well, I mean I think [snorts] an outside of the industry story that was similar was I I think I remember hearing, you know, like all the bottle caps are a lot thinner now than they used to be like on a on a water bottle. Like someone was just like, what if we can cut those in half? I mean again, it seems ridiculous on a one bottle scale, but then as far as for Coca-Cola or whoever else, like that's a huge savings over thousands or millions of water bottles or soda bottle, you know, it's just it's wild. Yeah, exactly. Yeah, we I remember I was working for Encana when I was a field engineer and we finally asked them why they wanted this 15-minute pressure drop watch at the end of each stage. Like, what are they using it for? And they're like, "Oh, it's just a nice to have." And I'm like, "You realize that's costing you a stage a day. Like, how much time we're sitting here watching the pressure drop? If you never using it, we can get a well done in, you know, a week faster or pad done in a week faster." And like, That's wild. Yeah. I did I'm thinking about that. When Bobby and I So, Bobby and I worked at RDS and we were working on new products and one of the ideas that I had was, you know, just automating the chemical process to it on the frac side, at least the logistical side of it, right? It's like, "We've got these totes. We know how much is in the tote when it arrives. We know how much we're using from the tote from from the LAS directly. And yet, you know, there's I've so many stories of just sitting there looking out my window and there's like 10 sand trucks because we screened out and no one told them. And so, they're charging us to merge now. And it's like, there's so many optimizations in [laughter] a frac side that which is cool cuz, you know, it's there's significant like real-world improvements that you can bring uh some some much more easier than others, but Yeah. Uh there are Our ability to put sand into a well like for the cost that we do it now is pretty amazing. Yeah. The volumes that we do and you look at the frac tickets now and it's not much different than it was when we're putting a tenth of the sand in the well, so we're seeing huge cost savings and it's all from stuff like that. Yeah. Well, then they got to they got rid of uh of pumping ceramics and resin coated and all the ex-I remember giving like I remember my cost for ceram-one I don't re-you don't remember whose it was. At one point my cost for ceramic were like a was a dollar a pound or something. Insane like that. And it's like I I can't even comprehend. You know, of course it was all engineering driven, right? We didn't know We didn't know better, but why not try? Like [laughter] it took EOG coming along pumping their leftover material for everyone to Well, that, yeah, you know, they didn't they didn't come out and say it, but it's like 100 mesh is a byproduct of 4070, which is what they were mining at their sand mine, and they just had a giant pile of it and they're like,"Hey, it's free."I guarantee you that's what happened. Like I have no idea cuz of course EOG doesn't talk about anything, but uh you know, it's like why not? Let's just see what happens. And sure sure enough permeability and porous or permeability of sand is greater than shale. Doesn't matter how how good the sand is. [laughter] Our rock is so terrible now it it almost doesn't matter what we're pumping down there. It's going to be better. Yep. It really I I feel like our industry is very good at overcomplicating things a lot of times and that is one of those that is absolutely true, you know, it's Yeah. We definitely over-engineered that for years. [laughter] Yeah. I mean, we went from a very driven like engineering driven industry to a more of an operations driven industry in the last 10 years. And it had to. That's the only way it's shale's competitive. Mhm. Um but we're more of a a manufacturing kind of mode at the moment than we are explorers you know, doing the technical work. We'll probably be explorers soon. You know, we'll have to find the next shale, but
So [clears throat], I'm not going to get into your background, but like we'll work back from like this most cuz this is your second stint with Liberty, correct? Mhm. Uh but talk about like so I mean you came in looks like 2021 as a data engineer and architect, but I I don't know if I'm not not really seeing much of that ahead of that. So, how did you dip into data ahead of that or And then also just kind of like you talk walk us through like the journey from you know, 2021 through to here cuz it looks like you were a big part of implementing a lot of different technologies at uh at Liberty. So, just kind of curious on your journey to the data side and then if you want to talk about some of the different decisions and why you chose things to implement and everything. Be pretty interesting. Yeah, I mean, yeah, it might not come through but Liberty is a very driven data-driven company. We have very technical upper managers. Um, they they make data-driven decisions. Not purely data off driven off of data but definitely data-influenced influenced decisions. So, we've always had a very strong kind of foundation in that at our company. Um, and when I was when we had the downturn in 2015, you know, we were looking for things to do. I decided maybe this is a great time to to learn some more programming languages. I learned some in college but kind of fell off as I became a field engineer. Um, so I started getting into using Python, using databases. Um, I kind of told IT what I was working on so they threw me a a database instance to let me kind of land some of my data. And really how it started it was I I built our quoting engine at Liberty. So, kind of how we get our KPIs and our analytics off of each quote. Um, and this data, similar to the you know, field engineering stuff was kind of sitting, you know, siloed in documents. Um, so I built a an XML process as well that kind of exported out that data and then databased it. Um, that led into this whole forecasting process that we built um, that kind of takes this data and then we did like a 3-month or so look out at uh, revenues and costs and it's you know, started off a little rough. You know, we were getting pretty close but not 100%accurate. So, then we started really diving into where those inaccuracies were, making it better. Um, and then we went public in 2000, I believe 17 or 18. Suddenly that became a a very big part of my job was making sure that that forecast was correct and we we drove a lot off of that. Um, the team that you know, I kind of started there uh, is still running that process today. So, we took it from kind of a document base to a Salesforce base. Took a lot of that logic, those those analytics and just kind of formalized it into an application. So, you know, we still use that same process today. So, that's kind of how I landed in data was it was more It's something I've always been interested in. You know, I probably should have been an electrical engineer not a mechanical engineer out of college cuz those were the classes I ended up like liking the most, but um Yeah, I kind of it I kind of fell into it, to be honest. It was something I really liked. There was opportunities at Liberty to make Liberty a little bit better through that, and so I started working a lot on it. Sweet. That's cool. So, and that was your first stint with Liberty, right? Cuz it looks like you were there 2014-2018. Uh yeah, I started in 2012. So, basically in I started out in as a a sales engineer or a salesperson, kind of moved into more of a sales engineering role. And then I took over the demand management group, which was that kind of forecasting Half of my role was forecasting, um building all that out, and helping the sales team on that. And then the other half was kind of a coordination role between different departments. So, we had pretty tremendous growth in that time, so we needed to make sure that sales, finance, operations were all kind of on the same page. So, that was kind of my role as kind of an intermediary between all those. Nice. So, all right. So, you got kind of started getting into the data and stuff there, and then we could talk about MassMutual and other stuff, but then you you came back it looks like 2021. Um so, can you talk about like some of the stuff that you got you got, you know, the initiatives that you guys had and, you know, maybe some of the tools you chose and why, and, you know, how it all fits with what you're talking about earlier. Were you Did you come in When was the Slumberjack acquisition? It was right after the Slumberjack acquisition. that's that's another thing I was going to ask you about was how not only did Bobby's question, but also how how how did that go as far as merging like yeah, integrating an M&A I feel like on the services side is probably even worse than on the operations [laughter] side, but It's Yeah, I don't know. I I suspect it's probably tough on both sides, but um Yeah, maybe I'll touch on Schlumberger real quick. Just uh talk about that. Yeah, the integration Schlumberger was was pretty good at helping us integrate. So, they got stock in our company was was how the deal was formulated. So, they were invested in making sure we were successful. Um so, from that standpoint, you know, it was a lot of work trying It was culturally was probably the biggest challenge. It's It's just different companies. Schlumberger is huge. And they've been around since the beginning of oil services. Um, and so, kind of transitioning from that, uh, over to Liberty which is kind of, you know, you know, we're a big company, but we still feel a little bit small compared to something like that, uh was a bit of a challenge, but I think it, you know, worked out pretty well in the end. We brought over quite a few of their people to kind of run some of the stuff. We were never a wireline company before that. So, that was kind of our first entry into wireline. Um it That opened up a lot of opportunities that we um had always kind of thought about peripherally, but you know, we we tried to stay focused on frac until it didn't make sense. And now it makes sense to kind of be that fully integrated service company. So, yeah, I'm not going to say it was easy, but it was a lot of work, but certainly worth it in the end, I would say. Um We definitely And we also got some of their uh you know, software through that that we kind of looked at, um things of that nature. A lot of what we ended up keeping what we had just easier that way, but we tried to plug in, you know, good ideas where we could. Easier to integrate theirs than try and rip and replace. Well, and theirs is a lot of custom-built applications. That's always [snorts] going to be tougher to integrate. Um But on the Liberty side, yeah, we started getting into more of um big data. You know, we've we were kind of siloed into maybe one or two databases. Now we're really expanding into other databases, other cloud infrastructures. We're We're primarily in Oracle and AWS, but we have some GCP, we have some Azure. So, how do you integrate all of those technologies? And when I came on board, my CIO is like, you know, we can teach you that stuff, or you can learn that stuff. It's harder to learn FRAC. Like, and so having kind of that background in the especially the history being here from the essentially the start of the company really helps kind of understand the challenges that we had. Um and so a lot of the projects that we've been working on recently is how do we get that data accessible quickly across the company. Um and we had been doing some of that through replication, and now we're trying to get away from that. How do we just expose that data from its current where where it's living right now, and then ingest that into where it needs to be? Um so that's been a lot of our initiatives. We've started uh we also started building some new infrastructure like Salesforce or sorry, not Salesforce, Snowflake. A lot of our customers already had Snowflake. Um we were looking for a platform to do data sharing on. Snowflake was ridiculously easy to get spun up. I think I think I had a POC running in about 2 hours, which is nuts for something of that scale. Um and so that was an easy win to kind of get that rolling. Um still going going today. We're trying to uh build that out and make it a little more prevalent in our infrastructure. Can Can you talk about that process real quick, just for people that may not be aware? Cuz I don't think we've talked much about it, but we've talked about different ways of data sharing, and you know, obviously people are familiar with REST APIs and you know, FTPs and you know, emailing CSVs. But like, I mean, like I used to I mean, you know, full disclosure that's kind of how we got this podcast set up. I mean like Liberty's doing that for for Grayson Mill and and again, can you just talk about what a what you need like three pieces of information from your client's Snowflake tenant and then Yeah. basically plug it in and their data's there for lack of a better term. Like Yeah, so what Snowflake built and is really truly ingenious is your data sits in your account. And then when you want to share it, we've set up a a view based security process, a secured view essentially, that if based on the client's account, we we expose their data, their segment of the data, to them in their account. So, they don't have to copy the data. It's still sitting in our account, but they're now able to access it access it across accounts. Um so, it you don't need any replication, you don't need APIs, you don't need any of that. Uh you just need to set it up on your side, which I think can be done in two or three clicks. It's pretty fast. Yeah, I've set it up before. It's stupid easy, [laughter] you know, quick to set up. That's probably the best best way to describe Snowflake. It's stupid easy. Like, they really simplified cloud databases. There's also no like transfer fees or any like right? Because it's living you know, the storage decoupled from the compute. So, like Liberty's technically paying for the storage, right? But then like we're just paying for the compute to access it. Yeah. But the And the upside to the client is it shows up in their database as if it's their data. They don't have to manage it, we manage it from our side. But then they can integrate it into any of the data that they already have in Snowflake. So, really opens up a lot of opportunities for our clients to quickly do analytics off of that instead of having to load it into their database and then try to find all the the ways to connect it together and things like that. It it really simplifies it. Yeah. And again, yeah, I mean, again, we're just writing SQL against this; way easier than having you need someone in house who can write some Python, and then they need to be able to make sure they keep it keep it alive, or do they need the retries, and then does data duplicate, you know, just And again, like I think we've talked about it here, too. People underestimate what it takes to serve up a REST API. Like, that's a whole other service that you need to keep up. You can have rate limiting and to have a good API with good documentation and then you've got versioning, you know, it's just it's a nightmare as well. So, something like this where you can just share it across this SQL database. And again like you said, Snowflake is such an easy button. Like, I mean even just turn to spin up an instance of if you like to test it out is just so trivial and everything just works. Um Yeah. It's uh it's pretty fascinating. Yeah, the only challenge with Snowflake is you have to be a little careful about how you set up your compute because you can have some cost drift there. But honestly, we haven't had that issue; as long as you're, you know, set up your warehouses correctly, it hasn't been a big issue. Yeah, no same here. Uh I mean Yeah, I've had some people it's so expensive. How are you guys I guess like dude, it's not that bad. I mean I quoted a number it's like oh, that's all it is. Like yeah, but if you you know, and I don't know if you saw there was a blog someone had it, but it was he's like I saved my company $500,000 with a one button click and it basically was just you know, but it took him 2 years to convince people that he needed the access and he went and turned and he just turned down the idle time to the 1 minute. Yeah. Yeah. Um and now you can actually go sub sub 1 minute now. Um but you know, but people the default was 5 or 10 minutes. So like you could run a 1 5-second query and then it would run for 5 minutes. Like Exactly. you just if you [clears throat] it's it's always about learning and understanding the technology you're using. Like you know how to use it and know what its limitations or you know, what the different levers to pull, you know, but it's it's can make your life a lot easier. Also, just going [clears throat] into it knowing that these are businesses set up to make money and maybe they're not going to always make it as easy to you know, it's not out of the box set up to make to be as cost efficient for you as it could be. So just understanding that I think cuz that's like I mean that happens all the time. Bobby and I have both had it where it's like oh, where's this random AWS charge coming from that I shouldn't have? It's like oh, I I left the database spun up that I was testing in and forgot about it. [laughter] It just happens. Yeah. [snorts] I think Snowflake is a little easier than some AWS like their Kubernetes service and you know, yeah, we'll spin it up and spin it down for you. The spin down you should be cautious though on the spin down. You know, maybe wait a little bit. So yeah, you have to be real careful about those things. Yeah, no for sure. Um so so you're talking about like something like the replication and that you try to move away from some of that. I mean, can you talk about cuz I I think I mean again, I see it on on your LinkedIn. You guys were using the same thing we're actually using a GME as Fivetran to replicate some data in there. Uh really great service. They can definitely be, you know, a little more on the expensive side, but you know, you pay for quality sometimes there. But, um, I mean, is that something is that where you were kind of hinting at, you may or may not be trying to, like, you know, look at where you could do less of that? Or, um, and then what has been your experience with with Fivetran? Fivetran's been great. You're right. It's it's another easy button, to be honest. Their service is like as far as replication setups go, it's one of the easier ones. Um we had that running in about 10 minutes. Like it's it doesn't it's so easy, it's fantastic. At least as long as you're using, you know, standard sources and destinations. Um it is expensive though, you're right. And the other issue that, you know, we've been running into with replication is um you lose you lose a little control over data quality. So, it's harder to maintain two data sets um and make sure that those two data sets line up. Um there's tools you can do, but then that adds even more cost because then you're running um queries and um compares against two different data sets. Um and so you we you know, especially when I first came into this role, we were running uh quite a few replications that were constantly going down. And so, we're fighting, you know, instead of spending time building new things, we're constantly doing maintenance on old things. Um so that's why we've tried to go away from replication. Um we've started looking at things like uh Iceberg formats. So, Iceberg has a really nice way of kind of consolidating data down into a very consumable size. Um and then exposing that across databases. So, instead of doing a copy over, you expose that data. And if you do need to write it to the the then you just do a database function that does it for you. Um things like that have really simplified how we do our replication. We still have replications running in our company. Um but we really try to minimize that where we can. And can you can you talk for people that may not be aware of what Iceberg is? I mean it's a it's a file format, correct? Essentially. It's a file format, yeah, that it's an Apache open source kind of format. Um it's starting to become more um accepted across uh platforms. It's kind of one of those uh data lake type um services that are coming up. Um the advantage that we've had like especially when it comes to like combining data, you can kind of combine it into um chunks. So if you have 1-second data packets coming in to an S3 bucket, you can pull that over into an Iceberg format that kind of consolidates those all into one. Um so you can kind of think of it similar to like indexing on a database um except that you're physically pulling those files together. Um and then a lot of databases are now like for example Snowflake is was one of actually the early movers um in accepting Iceberg formats. Um so our engineering data actually goes into an Iceberg format and then Snowflake views that um directly and then writes it to our database. So we have a function that runs every 5 minutes that writes it to the database and that's how we kind of keep it there. And the advantage of that is if you don't need to do like if if query performance isn't a big deal, you can just leave it in your Iceberg. So if it you know, it's not going to be as fast cuz it's going to become a large data store. Um but if you do need query performance, then we write it to our relational database where we have that SQL advantage. So have you guys messed around with like the Iceberg tables with Snowflake cuz I think you can query against your S3 buckets or blobs, you know, with the Snowflake compute. So basically you're managing that's decoupled storage, right? And then you're able to Yeah. you know, query it in line essentially. Yeah, that's that's essentially how we we built our new process. We were using Fivetran and Fivetran was replaced with this Iceberg read. So, it's it's living in an Iceberg table and then in Yeah, I can ingest it into a Snowflake. Sweet. Um and then also means that you guys are using Salesforce. Have you messed around cuz Salesforce has it now, too, where can't they do like that zero copy into Snowflake? We have not played too much around with that. Um Yeah, I do I've have seen it. We looked into it, but how we're running it now um has been working, so we haven't really messed with it too much. Got you. Cool. And then like so, is it is everything, you know, essentially going into Snowflake? Like, it sounded like you got your...you were talking about earlier, you got your ERP, you've got that telemetry, you've got...I mean, are you...are you trying to land all that, or even just, you know, versions of that in there and blend it across disciplines where you can? Or how... how's that all? Yeah, so we're we're primarily warehoused in Oracle, so we leverage Oracle's size, to be quite frank. Sure. Um where we can't outgrow it. Um we use Snowflake primarily for external shares, so uh both clients and then some of those vendors that you talked about, Well Data Labs and Coldbore and those folks. Um we've set up processes through Snowflake that kind of secure that. So, and that's the other part where we really like about Snowflake is it handled the security for us um and very well. Um and so, that's why we use Snowflake for mostly external sharing purposes. Uh internal kind of analytics and warehousing, we we mostly leverage AWS and Oracle for that. And are you just like using Oracle like relational databases for that or Mhm. Nice. Yeah, some of that. They also have uh they do have object storage is what they call it, similar to an S3 bucket. Um we do use some of that. Um and partly it's cuz we have a Oracle ERP, so it's just nice to be close um in that sense. Sweet. Um Yeah, cuz we haven't really talked too much about Oracle Cloud, but I mean I I think is their blob storage is it S3 compliant even? Yeah, you can actually read S3 buckets into Oracle. It's very similar format. I don't know if they're their object storage is is the exact same. Yeah, cuz I think, cuz I think GCP, obviously AWS I mean, I think Azure was the one I think for sure was not S3 compliant but I think there's a handful of, like, other object stores, you know, from the other vendors that are actually S3 compliant, so you can kind of treat them the same. Um Interesting. But uh Well, cool. Um And so, you know, anything What about like on the telemetry side? I mean like you talked about like so say, you know, you're going to be going to Australia where you might have 8 hours of coverage and stuff. Um I mean, are you guys on the IoT side? Are you getting down or messing around with like MQTT or different like messaging protocols? I was going to ask, are you doing stream, batch, MQTT? What's the Edge processing using Yeah, so we are both. Uh so, we do do MQT within the van itself. Um We also and I believe there's a messenger in there that then collects it and then streams it back to the cloud. So, we have a um we're actually using a TIBCO product. Um some so, it kind of integrates directly into Spotfire. Stream base? Mhm. Nice. So, we're using that product to stream the data out lands into a database that then we collect and then drop into Iceberg format for long-term storage. So, um that streaming data comes out in 1-second, you know, packets per channel. Um so, there's quite a bit of processing on the back side to kind of flatten that out. We pivot that data into column format and then bundle that all into a stage file. Nice. So, you said you're you're ultimately having to pivot it. So, like everything's just kind of coming in like kind of tall and skinny cuz you said it's one packet per um per stream. Yeah, per channel. So, I think one advantage you guys have in doing that cuz again I don't I'm not aware necessarily of any of the other competitors. I'm sure some of them are are starting to do it or wanting to do it, but yeah, before it's been at the mercy of, like, you know, again we were trying to do it at RDS, you know Well Data Labs are trying to do some of the streaming, Corva, Kelvin, you know, you name it. Obviously Colebourn is doing a lot of it but and you know, uh Rivet but but trying to like normalize or hey, come up with like a standard but then trying to like normalize that but then like you guys actually have control of the channels. Which is a huge benefit. I mean cuz are you guys still taking it from a serial stream ultimately or are you able to get it even kind of head of that in a better in a better format? Cuz I mean you guys see the CSV stream was kind of rough. Yeah, the CSV is very rough. Uh so, we do a mapping within the van itself. So, once it lands into the van, there's a mapping uh that the engineers set up. They should only have to set up once per pad or once through major change and that's the only time they have to touch it. Uh gives us a lot of advantages. One, it makes our streaming. So, once it's once it's through the filter, streaming's pretty much normalized. We don't have to worry too much about it. Um the other advantage is we used to have to do a lot of mapping with the the the third party companies that are also doing the streaming. Mhm. Um it was quite time-consuming in the field. So, we're trying to eliminate that by just normalizing everything. So, they just plug into the channels. They pull out the data the same way. So, no matter what van they're in, uh it should be the same channels coming back at them or it'll be named the same. Um so, it makes their side even easier as well. So, um there's hopefully some cost savings and some time a lot of time savings ultimately uh going to that method. Also, everyone else that's fracking in the world right now, please standardize your naming convention so that everyone's life is infinitely easier because [laughter] it never ceases to amaze me. Um There's that third party group that's trying to do that. I'm blanking on the name right now. They made some headways in the What's that? Was it a rivet? No. Um it's like a inner It's an independent group. Uh I'm blanking. It's a an acronym. I can't remember right now. They made some headways in the geophysical space cuz those guys all played pretty nice together and just figured it out. But the completion side got some some big personalities who don't want to change anything. So yeah, there's a little bit of challenges on that side. No, well that was the thing is it's like I remember we probably told the story before but Bobby and I met with a guy at an operator. They hired him to bring to do a bunch of fracking analytics and stuff. And the first thing he had to do was write a machine learning model to normalize the naming all the headers of all the decade of historic frac data that they had. And it's like yeah, that's a huge problem [laughter] because no one was thinking about how important you know, calling even just the chemical names the same thing, right? Like it's just Chemical names, even pressure. Yeah. Inline pressure, surface pressure, treating pressure. That's probably more than I don't even know about. Yeah. So which one is it? And what what one do we care about? Yeah, no it's uh it's one of those things until you get into it you just have no clue how messy it [laughter] really is. Yeah, that legacy data is yeah, it's very messy. But no, I'm all for it. I would love it if we all could just start naming things the same but there's like everyone's life either. There's no reason. Okay, we don't have to agree on a format. We don't have to agree on a frequency. We can just but we can as an industry come up with what are we going to call treating pressure? What are we going to call like that is the simplest thing in it and no one has done it yet. And it's just like just put out a blog post. Here's what we're using. This is what we're going to use. And then if you want to use it, great. But like yeah, there needs to be you know, open source community-driven standard that everyone can use because everyone does it. We talked about it it before. It's too bad though but I mean even they were trying to get the XML for frac which again I'm sound like Brian or the wits ml rather, but Brian sounds like you're a lot more uh familiar with XML or don't mind working with it. I I hate working with XML personally. Um But um but it again, it's one of those things like it's maybe a Stockholm syndrome, but I mean, at least on the drilling side, I think that's where a lot of that growth in the on the analytics side came from cuz they had a standardized format that people knew how to tap into and use that Yeah. people could build off consistently. You know, if it's like,"Well, I've got to build this process if it's a Liberty van. I've got to build this process if it's a Halliburton van."and then, you know, it goes if you do you know, a different crew on a different side of the country. going to say, it's not even just the van. It's even the crew, right? Like Yeah. Yeah. It's uh yeah, it's crazy. But yeah, that's I I'm adamant that that's why drilling is probably 10 years ahead of of completions on the data analytics automation side because hey, it's a hell of a lot easier to build uh any kind of model if all the data is standardized, normalized with the same conventions. Like it it's not it's not rocket science, but No, it's a challenge that yeah, I don't I don't know if we're going to figure out anytime soon. I hope we do, but I'm glad you guys you've been to you guys have been to frac conferences, there's literally never agreement on anything at those, especially on paper talks. I don't know if we'll ever agree on this one. No. Just glad some someone in the industry like you guys is at least doing that internally, right? Like just And you know, ultimately I see that as the these are things that in my mind that start to become advantages, like selling points almost over the long term because everything is pushing towards, you know, leveraging data, data-driven decisions. We've been doing that forever. And yet, it doesn't it hadn't seem like many people you know, you had Corva, Coldbore, RDS, Rivet, all you have all these people coming in to say like,"Hey, we're going to solve that problem."And like,"Well, it's a hell of a lot easier for Liberty to solve that because it's their stuff, right?" Yeah. doesn't make sense in my mind to have a third-party doing it when you guys have like it's your engineer that's setting it up. He's setting up the tables. He knows the channel mapping. It's like you throw more the more people you involve in it, the harder and more complicated it's going to get. But on the earlier you start on it, the cost savings just trickle downhill real fast. Like if if you start with a solid foundation, everything becomes easier and then you get to the fun stuff, not the boring horrible data engineering side of things. Yeah, you're not you're not blowing hours and days and weeks on build building language models to normalize the the headers for the prolific data. It's crazy. Yeah. It is. Well, one thing too and and again, I'm not I mean I'm not speaking for our procurement side at all cuz I I don't I don't touch that. But like I got to imagine and you know, you and John have both been on the sales side of this, too, but like being able to provide that this data service around the frack makes you guys a lot stickier, too. I mean, like, and yeah, I mean, obviously that I think there's differentiators, you know, as far as, you know, quality of the service and the equipment, everything. But it sounds like that's all a function, or, you know, in some way a function of having and using the data to, to be, you know, more proactive than, than some of your other competitors, probably. But again, if I'm able to get a consistent, solid data set from this company and I don't have to you outsource it to a a third party to get a good quality data stream, um, I mean, it's going to make it a lot harder to move off it, you know, when you got some processes [snorts] and stuff built around around that as well. So, I think it's just smart from the business side. Yeah, we're always looking for that easy button for our clients. Like how do we just make it better and faster and easier for them. And that's always good for us. Yeah.
So, I mean, like if we're going to work backwards then it so it sounds like I mean, you got into data with with you know, Liberty probably back you know, 2015. And it's not like you had done a little bit in college, but this Can you tell me about some of your other stops and you know, kind of how some of that data those data skills and stuff have translated? Cuz we've talked a lot about too like breaking into data and that sometimes it's not necessarily being in a data role. Like I, I think with the last couple years is your first time truly in a uh a quote unquote data role, but you know, just how anyone I think in in any you know, discipline can start leveraging data and uh and uh sometimes that's the most important, you know, where you have people that have that domain knowledge and Yeah. Yeah, I'm a from the ground floor. I'm a big proponent of that. Like it's easy In my mind, it is so much easier to teach someone the data side, the stats, whatever it may be, than it is to train them on the years of field experience that you get by sitting in the field. You cannot It is so hard to train people on that stuff. Yeah. Yeah, I think, you know, when when people are looking to get into I totally agree. Learn the domain knowledge of your industry first, no matter what. Um early in my career I was pretty fortunate someone recommend they're just like stay in the field longer than you think you need to. Um it'll pay dividends down the road. And it really did. I mean, I I ran um between being a a frac engineer and a project manager, I was in the field for about 6 years before I came to the office. So, you get all the little experience you get out there um for example, you're talking about sand deliveries and all that stuff. You see the issues that happen out there. You start seeing why this is a big deal. You see where it's coming from. You learn the whole supply chain around that just sitting in the field watching it, seeing those problems develop. Um and that's how you get to be a better um you know, contributor to the business down the road. Um and then let's start learning the data skills or learn them in in conjuncture um because there's a thousand online classes learn how to do SQL. There's a thousand online classes how to do cloud infrastructure. There's all of that that part is way more accessible than especially in oil and gas learning the skills out in the field. Um and I I think the other thing, I mean, if you're truly passionate about data, um then start looking for side projects, whether in your company or or on your own. Uh you can spin up free infrastructure in AWS or Snowflake or all these other places, Python is free on your computer. You can interact with that data and you can start learning how all that works. Um, the barriers to entry there are so small. Um, and you'll and like I said, I kind of fell into it because I saw a project I saw a data being siloed in Excel spreadsheets and said that's we probably shouldn't do that. How do I fix that? So, I spent a couple weeks and I figured out a process to fix that. And yeah, I was kind of real basic and my code looks terrible and I've I'm almost I would be embarrassed to show it to anyone, but it worked. And now we have a whole salesforce process that's very formalized and it works with our ERP system and there's a whole bunch of things that got built from that simple prototype that got the the ball started. So, and the best way to learn it is to jump into a project. You just I if and I'd I'd say on the data side, I the best data people I know know SQL. Um, it's not going away anytime soon. And it's surprising how many people don't know how to do it. Um, some people do it through the GUIs and that's a great place to start. Um, but if you really want to be a power user, you need to learn how to code it and it's probably the easiest language out there to learn. It's very easy to read. It it's really easy to run. Um, so things like that. Um, yeah, I took a couple online classes like I said during the 2015 slowdown when we're all sitting around hope you know, hoping for more work shows up while the Saudis are fighting us over oil prices. Um, so I took that time took that opportunity to learn some new skills. Um, and it really paid dividends on the road. Yeah, I think the big thing there is just just starting, right? Like it doesn't have like you said, it doesn't have to be pretty, it don't have to be sexy, it don't have to be using the latest greatest new thing, but if it's solving the problem, it doesn't no one gives a [ __ ] how the sausage is made. They care what it tastes like, you know? It's Yeah. And I think the other thing to keep in mind is it's okay to have a really bad first project. Like Mhm. like I said, I would not show you guys my Python code cuz it's embarrassingly bad. But, it led to probably 10 or 15 iterations later, it's much more it became much more formalized, and then we threw the whole thing out and took it to Salesforce. Um so, it's okay to build something that is kind of ugly, and it barely works, and it's cobbled together with duct tape. Whatever, it worked. And then, now you have the prototype to build something better on your next iteration. And I think that's one thing that people are a little scared of is the failure of your project or a failure of something, and it's [snorts] you know, it's not a bad thing. Um and it's not a bad thing if it barely works. Like, at least you got started, the ball is rolling, and now you have something to work off of. You made you learned some lessons that you'll take to the next iteration. Move forward quick. Yeah. They're iterations, not failures. Absolutely. Well, that and I mean, like again, I I that's the natural progression of any code or you can just like, you know, what? Make it work, make it work well, make it work fast, right? I mean, like we oftentimes you never even have to get that last part. It's probably fast enough cuz you know, you run a 10-second script is probably sufficient, and it's way faster than like the 30 minutes someone was taking each month to do that process or whatever it was, you know. Um I think the one thing I learned real quick, too, once I started coding was like the reproducibility or the reus- reusability of that. Like, all right, I've done this problem once. You invariably have a similar problem 2 months later, and now I've got a starting point. So, now I'm able to implement that faster, too. And I think that's the thing, too, where we can talk about speed. Is it developer speed, or is it the speed of the code? And for me, it's always a lot more about developer speed is more important. Again, if it runs in 5 seconds or 20 seconds, depending on the process, who cares? But like, if I can ship a product or ship some code faster the next time, and get a solution in someone's hand faster, I think, you know, it's a big deal. And I think that's where people, like, say if they get stuck in that Excel world, and like,"Well, I copy this CSV from here, and I do this "and they don't even write a macro to do it, you know"and I move these five columns, and I run a pivot table."Like, no, like if we can write that as code, we do that once and then we can just run that, you know, let the machine do it. Um yeah, and just all that kind of stuff is super powerful. That's the case for Python, honestly. I mean, there's way faster languages out there, but you can throw someone at a problem, there's probably a library that solves it, and you can have your code working [snorts] in an afternoon, where you didn't have to drive, you know, write drivers or do all this other stuff [laughter]. And now with I don't know if you're using GitHub Copilot or obviously ChatGPT, I mean, like stupid little things you want to put together, I mean, it takes a an afternoon into a 45-minute project or, you know, 15. It just depends on what you're trying to do. Like, it's pretty wild. I've got a It doesn't It doesn't come out perfect, but man, it gives you the framework really fast. Yeah. Yeah. I've got a OneNote uh whatever they're called sheet group uh of just Power BI functions that I use all the time so that I don't forget them because Yep. I come in and out of Power BI so often and infrequently a lot of the times, and it's like, I know I'm going to need to use this at some point and or I've used it multiple times in the past, so I'm just going to save it here with boom. Yeah [laughter], it's It's the same thing with code, though. It's It really is. Yeah, and like you you guys said, I mean, all the libraries out there just make things so much faster. [laughter] One of the funny things I do on occasion is I'll Google a problem and I'll see a a purple link that I was like, [laughter] That was it. Had that problem 4 months ago.
Well, that's a funny one, too. And I I don't think he he he didn't bring it up. We we interviewed this guy JD Long. I don't know if you've ever seen him on social media or not. He's pretty big in the R and Python communities, but I think he's mentioned multiple times looking up a problem on Stack Overflow and realizing he answered it like 2 years before. [laughter]
Yeah, I mean, people joke like,"Oh, you had to Google it."But, I don't want to remember all this stuff. I don't need to remember all this stuff. It's It's there. I can As long as I can get to it, that's all that's important. Yep. 100%. Um well, great. John, do you want to jump in Uh, we'll into the lightning round or Yeah. That that hour flew by, Brian. That was that was good stuff, man. So, we like to end it with lightning round. We'll just pepper you with short short Yeah, couple quick short answer questions, or whatever. Um, cool. My go-to one lately has been favorite uh video or board game. If you're into either of those. You can also say you're not into either of those, so. No, I mean, I have little kids. That's my board game life. Those are my favorites right [laughter] now, so. Uh, you know, video game that I always kind of fall back to, I'd say it's I hate to say it, but Call of Duty. I just especially the older ones. I just love those. And I like it cuz I can jump in for 15 minutes, play, and then jump out, and then I'm like, all right, I'm moving on my next thing. Yeah, I got that. Scratch the itch, and uh I'm good. I'm done. Yeah, I'm not And then some kid kills me, and I'm like, all right, I'm good [laughter]. Move on to the next. All right, now I have to go mow the lawn, and actually do something. Yeah. Yep. Um, all right, so you mentioned uh Python libraries. What's your what's your favorite open source library? Man, um It doesn't have to be favorite. It can be just useful ones that you think other people might use. I mean, I'm going to be honest, the standard libraries are the ones that I use all the time. NumPy. There's some uh the SQL um uh Alchemy. Um, I use it quite a bit. Um, and then a lot of times the develop like the owners of the um databases or the connectors usually have their own libraries, and it's fantastic. Like Snow and I think that's one of the powers of some of these bigger guys. Snowflake's connector is fantastic. I had a Python connected in, you know, 5 minutes. What's wild too, their integration with with Pandas. Like you you can have the Snowflake Pandas thing, and you go right from the two SQL, and then just, you know, goes works seamlessly. That's pretty impressive. I guess yeah, Pandas is probably the other one. Although, I don't I haven't used Pandas quite as much as I used to. Mostly cuz I'm doing data processing or any analytics anymore. Yep.
[clears throat] Favorite um Just some some interesting folks that you you follow either on LinkedIn, Twitter, wherever you may be that other folks might find interesting. Uh Carly Taylor, I I've known her for quite a while. She's over at Call of Duty or at Activewear. I think it's the name of the company, the parent company. She's pretty tight with Jessica Iriarte, right? Yeah, I had lunch with her two weeks ago with Jessica. So I follow her as well. I'm trying to think of some of the other ones. Uh You know, I think Adam Grant is another writer I really like. He's just more kind of general business kind of he throws nice thoughts out there that make you kind of sit there for a minute and and work through it. I've always liked him a lot, too. And then He's got a really good book, too, doesn't he? I'm trying to remember what the name of it is. Uh yeah, I'm trying to remember which one his was. I read it two or three years ago, which is why I started following him. Yeah, same. And then [snorts] Malcolm Gladwell is another person I tend to follow quite a bit.
Was it Originals? I think there was that. That's the one I read of his, but he's got a lot of good books, though. Yep. Yeah, the other book that I read recently I really liked was Thinking Fast and Slow Thinking Fast Slow. Um I'm blanking on the author's name right now. I read that about two or three years ago, but I really kind of changed my perspective of of attacking problems and how I try to avoid biases in that. Yeah, that's a good one. Well, one of the questions we normally ask is about books. So you got you checked that one off right there, as well. Two birds with one stone. Yeah, yeah, there you go. [laughter] You want to wrap wrap us up, Bobs? Yeah, um All right, last one we'll do favorite place to go on vacation. Oof, that's a great question. you you live in Denver, so you got uh You got it in your back backyard vacation all the time, right? Uh probably the mountains, I'll be honest. I love the mountains. I grew up here in the mountains. I go up there all the time. I do some larger bike rides, group rides every year up there. And you know, whether it's mountain biking and then in the wintertime I love skiing. Try to ski as much as I can in the winter. Trying to get my kids into skiing so I can continue to ski. [laughter] I need I need to do that with golf. Yeah, exactly. I love golf too, but man I man if it's between that and the mountains I usually end up up there. But yeah, if I'm traveling Probably probably the beach is the other place. It's nice to just go to a place where you can't really take your phone and you just sit there for a while. Yeah. Yep. Awesome man. Brian, we appreciate you coming on. It's been great. Yeah, thanks you guys. It was awesome. Thanks for your time.