Hypercube Podcast Transcript

Title: Data Innovations for Oil and Gas and the Importance of End-user Testing

Host: Adam Sroka
Guest: Jessica Iriarte

Intro: Welcome to the Hypercube podcast, where we explore the cutting edge of data and AI in the energy and utility sectors. I’m your host, Adam Sroka, founder of Hypercube, a strategic technology consultancy, supporting asset owner operators, traders, route to market providers, and energy services companies unlock the power of data and AI.

Join me as I sit down with experts from across the energy industry to uncover how they are navigating today’s data driven challenges. And without further ado, let’s get into today’s episode.

Adam Sroka:  Hello and welcome back to the Hypercube podcast. Today I’m joined by Jessica Iriarte, General Manager for Completions at Corva. Jessica, how are you? 

Jessica Iriarte: I’m doing very well. How are you doing? 

Adam Sroka: I’m very good. Very good. Thank you for coming on the show. I guess for the audience and for those that don’t know you or Corva, would you like to just introduce yourself properly? 

Jessica Iriarte: Yes. I’m the General Manager of Completions at Corva, basically overseeing the product and commercial strategy for the Completions product line.

That means doing a lot of different things every day, working across different teams. with marketing and product and operations and sales, basically a strategy. So it’s quite a diverse and fun role. 

Adam Sroka: We had a chat a while back and I think, yeah, that undersells it a little bit. Like I was so excited by some of the use cases and like clever big brain stuff that you told us about last time we talked.

Keen. Yeah, definitely keen to get into that. But before we do, how did you get here? Like how did. Your career developed to the point you’re at now? 

Jessica Iriarte: It’s quite interesting. Like I, I’m a petroleum engineer, undergrad and grad and I grew up in an oil city, so that kind of was in my path I guess.

And I started working in the field offshore in South America. And then I worked in Europe for a little bit and then I moved to the US to do my masters and I guess it, I found tech in the US and data and then started to gravitate towards. That world still within the old industry. I worked a little bit outside on the industry, but I think it’s been like a journey of where I’m at today is thanks to, having that field experience, having worked in tech and software and how they work in data science for time series, like sensor data, which is like very specific to the work we do here.

So I do think it all built up to this point. 

Adam Sroka: Yeah. Okay. Yeah. The way you do Corvua, can you expand on that a little bit? What problems are they trying to solve or what are they trying to achieve? 

Jessica Iriarte: Yeah. So Corvua is the way I like to describe it is like a real time insights platform for well construction and lifecycle operations.

Basically we collect that real time sensor data from drilling rigs or from track fleets and provide visualizations in a way to look at these data with dashboards that you can build. We have, we’re also like our slogan is the app store for the energy sector. So basically we have all these different applications that you can add to your own dashboard and then you can combine them, if you’re monitoring drilling and collisions.

We also have geoproducts like geocompletions or geodrilling. So those are really great collaboration tools because then you can have, the geologist, the driller, the completions engineer, and maybe other people involved in these operations looking at the same data, one source of truth, which is really hard to get, especially when it comes to data integration in the industry.

Adam Sroka: Yeah, because I guess of all, of like, All the industries that I’ve spoken to, I think drilling especially is very like forward thinking and quite far advanced when it comes to sensors and like the volume of data that is produced and the variety. Do you find that to be the case? It’s quite forward like in front of other domains.

Jessica Iriarte: I think it’s the most standardized for sure. When it comes to data collection and formats, call patients, we’re still getting there. But when it comes to sensors, I think both operations have a lot of sensors and a lot of equipment. It really depends on the type of operation. Sometimes, obviously, offshore operations are the most technologically like forward, usually, they are the most expensive and more like risk, operations as well.

Adam Sroka: And producing dashboards for this domain, non trivial, because in my experience, again, the people, the experts, the domain understanding comes from technical, really clever folks that know what they want, but maybe don’t I don’t quite know how to translate it into what you’ve got. That’s quite a hard job bridging the gap.

I guess in like a product world as well, that must be really quite a difficult challenge to, to be that translator between the kind of two worlds. 

Jessica Iriarte: Yeah, that is true. Our users are very diverse. We have people that are in the field and looking at, have Corva in, their their rack van or their money train van and are looking at Corva or we have real time remote centers.

That are in the cities and the main cities that where customers are monitoring those remote operations from different fields. So the technical and operational background really varies and the tech savviness also varies. So it is quite challenging actually to translate, some of these operational insights back to them and why they should care about them.

Adam Sroka: Yeah. I always find that like super smart people that. are technical, but in another, like they’re not data or computer technical, but they are technical. I always find them often to be some of the most difficult people to help because they have, they’re not really mathsy and they’re really clever, but they don’t quite know the art of the possible or the way things are aligned.

And it could be quite tricky to satisfy them because they’ve got such a strong vision of what they need and how they might work. 

Jessica Iriarte: There’s a little bit of stubbornness to live too, right? Like you have your ways to look at data or to look at your own operations or monitor things, then moving away from that into a more like custom built sort of purpose built.

Sometimes it’s tricky. 

Adam Sroka: So it sounds like Corva is quite a complex product. What would you say are like the biggest data challenges and technology challenges that you have? 

Jessica Iriarte: We still deal with collecting massive data from sensors, right? These are remote locations. We use, We use Starlink in some of these locations to get our data in real time. In some locations, there’s internet, so it’s not that terrible. But, and then Corva itself. So we started as a drilling platform and then it has built into, then completions, GEO. Then adding what we call our Dev Center, which is basically people can deploy their own apps, so if they really want this very customized solution, they can, we can build it together or they can build it themselves.

And then we have Fusion, which is our like data quality layer. So it’s basically real time monitoring. We have like data observability, but we also have a like products of data observability, but then we have a 24/7 Ops team that is monitoring also data coming in. I think we have a really good architecture and structure to deal with these challenges, but it’s still challenging is that you’re still getting so much data.

You’re still going to get issues with the data lacking or different formats. I would say another challenge is our operations are constantly changing. And so we built solutions for the current operations and then they start changing how, even, Entirely how the operation works and how that affects how the data is produced and what it means, right?

So a big one for us right now in completions is SimulFrag or TrimulFrag. So basically you’re stimulating two or three walls at the same time. And we’ve had to basically refactor a lot of our calculations, a lot of how we stream the data from some of our core applications. To be able to adapt to those, new operations and new ways with that.

I think it comes like that funny thing that some calculations that you think are so straightforward, right? Pump hours per day should be straightforward, but it’s way more complicated. I always see. People only been talking about that joke of that question of how many active users do you have?

And it’s Oh, it should be such a straightforward question to answer, but it is not. After all, still, we have those in the industry as well. I know, 

Adam Sroka: I totally get it. Where it’s pump hours of days. So what do you mean by hour? Can we define hour? Can we define day? Like how, yeah, you get into the weeds.

Jessica Iriarte: When does your day start? Does it start at midnight at 6am? When is your morning meeting? 

Adam Sroka: What about leap days? Daylight savings? How do you do that? 

Jessica Iriarte: Oh, yeah. Daylight savings is a funny one. 

Adam Sroka: On the flip side of challenges, if you had to say one thing, Corvo was the best out, the tech team was the best out, like what would it be?

Jessica Iriarte: I think that user interface where, how we display these insights to our users is one of, I think, our big, biggest wins. Again, people are so focused in operations and like getting them, Interested in looking at this data and engage into why, how you can be, make your operations better or faster or more efficient or reduce whatever is your goal using these dashboards in this data.

It’s, it can be challenging, but when you have the right interface, it looks good. It, it translates back to the user. I would say that’s definitely our biggest win, I think. A lot of people can build a lot of data solutions, but then if you don’t really get people to actually use it, then they’re there.

I think the other one, and they are competing because obviously it’s two different, a lot of it, two different audiences are, data science, I think we’re definitely pushing the industry there. So that’s a differentiator. And on 

Adam Sroka: that end user testing, you shared some thoughts fairly recently about it.

Domain experts being like at the forefront of your testing experience. Can you expand on that? 

Jessica Iriarte: Working in software is interesting because you have a whole team of QA, right? But I’ve never seen that necessarily in like data science. Like you don’t have QA people that are constantly testing models.

It’s usually like the same data science team. Like on Boro you have some, like you wait for user feedback. And machine learning success really relies heavily on testing. I would say building the data set and then testing the model is the most time consuming parts, right? Building the model, deploying the model is, can be a little challenging depending on your setup, but really I think building a data set and testing it are the two, Biggest constraints and with our industry specifically where, like you said, all the data and all the operations are so specific to a domain where, especially when we’re collecting field data set, if you don’t have experience looking at the data or working in those operations, understanding how those operations translate to that behavior in the data, then you’re not, you can’t really build models that relate to the physical reality of what’s happening.

So I think that’s like a technical complexity that have this data science projects in technical industries like ours, or like science projects or operational driven, like sort of models where Okay. Again, you’re trying to predict a pressure, but if you don’t understand how that pressure currently behaves, what is affected by, why a change in that pressure is really not meaning anything because something else is affecting and it’s very obvious to the person that’s wishing to feel or on like works with these data all the time.

From an operational point of view. And usually if you are a data scientist that has no experience in the oil industry, like that won’t, you won’t know all of these nuances, right? So I think that’s where that testing aspect is so important because. That’s where you get usually, and there’s a Corbyn where I worked, where we build this kind of models.

We have a subject matter expert in, and I’ve, what I’ve seen also our customers do, right? You have subject matter experts very involved in the process. So these models can actually relate to the, again, the physical behaviors that are happening. in the real world beyond just patterns in the data.

I think that, yeah, again, it’s just so important to make these models be real. 

Adam Sroka: I think it’s quite common for people that don’t have the domain experience to, in my, in data science especially, to enter a domain with a load of really great skills and technology, point machine learning at a problem and a big data set, and then they start deriving insights or predictions.

And then, actually spend a lot of computational power and like model effort on getting the machine learning model to learn something that is actually described by like a fundamental relationship, like a physical relationship or some equation that is known in the field. And I’ve seen this happen a few times.

And so you get the illusion of like insight and value. But you’ve actually just really inefficiently rediscovered something that is known like in a kind of industry textbook and perversely, I’ve seen like senior stakeholders actually drive projects down that road and say, Oh, no, we want the machine learning to teach us about the domain and so we don’t want to tell it anything.

I was involved in a project where a super major funded this huge project and they were really adamant that we didn’t teach, like we didn’t use any heuristics, fundamental models, things like that. And eventually another senior stakeholder from somewhere else in the business stepped in and said, this is absurd because we’re wasting all this effort trying to rediscover like these fundamental understandings that we have.

Actually, we can just shortcut to the bit where the machine only derives real value. 

Jessica Iriarte: Interesting. So we’re working on a project right now and we, the model does really well at recognizing behaviors, right? But then a person looks like a subject matter expert looks at it and they’re like yes, that is happening, but that’s not the kind of behavior I want to look at because that’s obviously, Something that happened in the surface that is, it’s not necessarily something I care about.

That’s not something I can control in the future, right? Like pumps went down or the power went off. If you have an electric fleet or I don’t know, there was a storm coming. So they had to shut down everything, whatever. There was a leakage versus something coming from the subsurface, right?

Like your geo, like wherever these well is on there, whatever orientation of the well, it’s all dead. The angle that is landed within that specific, stage that you’re stimulating and how that is affecting that pressure behavior in the surface is what we’re trying to extract, trying to understand.

Again, the model, unless we like really introduce some of this kind of obvious physical relationships that we understand, it will not understand, how to differentiate from one to the other, or it will state the obvious a lot of times. So it’s yeah, 

Adam Sroka: We already knew that. As General Manager completions, it’s an interesting job title.

What does your day to day look like for us? 

Jessica Iriarte: It varies a lot, again, because I interact with so many different teams and also with our clients, my week, my day can vary quite a bit. I do try to have some sort of focus per week of okay, this week I’m going to focus on, I don’t know, partnering with sales and that’s going to be my focus.

And I’m going to be, meeting with customers, getting user feedback or understanding, things from the sales and this long term strategy kind of thing. Some weeks I can really focus on things and some weeks are just a lot more tactical where there is, things happening with our current operations or things are changing or things are breaking and it’s a lot of just supporting really all of our teams.

And then sometimes it’s really focused on like long term strategy. Where do you, do we really want to go? We see these trends coming, we see these trend changes. Do we want to enter this? Do we think, a lot of what I’ve been thinking right now, it’s like our, we do have a very established.

I would say establish AI product in drilling. So we’re trying to bring something similar to completions. And so thinking about where is the right entry point and how do we grow there or the right partners to grow there. But going back to the day to day, like it’s a little bit of a combination of all right, like again, supporting all of our teams.

Listen to our customers are saying a lot of times just spending the afternoon clicking around and looking at dashboards and see what some of our customers dashboards are looking like. What are they, using or not, and then partner, with our marketing team and as well, attend conferences.

So yeah, it’s. It’s a very diverse. 

Adam Sroka: Really cool. And you mentioned that you’ve got an AI product, you’re attending conferences, you’re keeping scope of what’s out in the market, you’re looking at like end users and how they work. It feels like you’ve got a good handle on driving innovation in what is otherwise seen as quite a risk averse industry.

Do you find friction to that? And how do you like, how do you go about that and build that culture of innovation? 

Jessica Iriarte: That’s a good question. I think that. Yes, the industry is risk adverse, but it’s not necessarily innovation adverse. So we are a very, obviously highly regulated industry. We do have a lot of constraints that we work with.

But the industry is constantly innovating. Usually what is more so in the operational and tool side, like in the hardware side. However, it’s started to shift into also like more automation and software and a little bit of data driven solutions. But the industry has gone from drilling just vertical, shallow wells to then drilling like deeper wells to offshore, to deep water, to unconventional reservoirs, like horizontal wells.

Now we have four mile laterals, this is six and a half kilometers long. Now have U shaped wells, that one like was really surprising to me. And now, we’re stimulating two, three wells at the same time. So when it comes to I think operational and hardware innovation, we really are an industry that thrives in that.

I think that, When it comes to data, there, we’re still working towards getting that trust from people because it’s similar to what we were discussing before. A lot of our users or a lot of people in the industry either have a very operational background, so their thought process is very empirical on their experience, which kind of becomes some sort of art and science in, in, in one and it’s actually really cool.

And then you have the very, engineer background where like traditional engineer background where. Everything it’s thought from like the physics base and, physics based simulation or modeling. So I think breaking a little bit of that barrier of not thinking of data driven solutions as a competitor to physics based modeling is one thing.

So I think when thinking about innovation, obviously we want to think about What’s new? What’s going to be the problem of the future and how we can solve it? But also, how can we know the way we want to get there. How can we start creating trust with ourselves, with those sort of solutions today with our customers and in the industry?

Adam Sroka: So that’s an interesting take. Oil and gas has this perception externally, but being quite high pressure there’s a lot needs to be done to stay competitive, driven by like oil prices and stuff. Do you feel that in, in your role and if so, like, how do you lead data and AI teams in that kind of environment?

Jessica Iriarte: Yeah, that’s a great question. I’ve never been asked that before. It is a high pressure. It’s a 24/7 industry, right? And I think for me personally, I started as a field engineer working 24/7 offshore. Living with the boat and then never knowing when you’re going to come back. And so you get used to that sort of Russian environment of moving fast and always on.

Not to mean that you don’t have work life balance, it’s like when you are like at work it’s. Usually it is an industry that it never turns off. Resilience, obviously it’s one that comes to mind right away. Being adaptable is very important. It is a high pressure environment, but again, it is constantly moving.

So being willing to change and not being stuck in your ways, especially when you’re working in software and in technology, that is so important. Staying curious is so important for innovation. And I think that sometimes when you are under pressure, that’s something that we forget and we default to problem solving and doing whatever is easy at the fastest.

But staying curious, questioning, is this really the right way? Are we really going the right direction? I think when you’re dealing with high pressure, like good, clear communication is so important as well. And I think personally with the people that I work with, or when I have a team reporting to me, something that’s really important is creating a culture where people feel empowered to take ownership of their work, especially against when we’re moving fast and we need everybody to pitch in and then be confident, that their contributions matter, that their voice is heard. 

Adam Sroka: And how do you think your international experience has shaped your sort of leadership style and what you bring to data and software roles? 

Jessica Iriarte: That’s a great question. I really love that you, your questions have been so different from any other interview that I’ve been in or a conversation.

And it’s been quite fun for me, just as a side note, sorry. The two things that stay now, is perception. It gives you so much perception, like it opens your mind to many other points of view and types of operations and types of dealing with situations, types of constraints, practices, like you really get a lot more exposure.

And there is no other way that I could describe it, I think, that, that is the main thing that I’ve gained from working in different places, learning to adapt, to different cultures, to speak in different languages, being open minded to, to a lot of different situations. So I think that I definitely take with me because it always helps me like not to It also always helps question my biases and, being in that way, open minded to not just question other people’s opinions and biases, but mine alone.

Adam Sroka: I guess. Yeah. Final question is always. If people want to find out more about you or Corva, or if there’s something you’d like to promote or plug, where should we be sending them? 

Jessica Iriarte: Definitely my LinkedIn, my personal LinkedIn, and then Corva’s LinkedIn. They’re always, we’re always posting there, what we’re up to, where we are in the world, and then our website corva.ai is where we have a lot of our in use products and case studies and in use. 

Adam Sroka: Amazing. Thank you very much. Really interesting conversation. And yeah, I wish you all the best in the high pressure world of autonomous drilling. 

Jessica Iriarte: Thank you so much, Adam. I really enjoyed this conversation. Thank you so much for being creative with the questions and for making it feel so conversational. It really It was a lot of fun. 

Outro: Thank you for tuning in to the Hypercube podcast. If you enjoyed today’s episode and want to dive deeper into the world of data and AI in the energy sector, be sure to subscribe so you never miss an episode. We’re always eager to hear from our listeners. So if you have any questions, any topics you’d like us to cover or feedback, you can reach out to us on LinkedIn or check out our website at wearehypercube.com. And remember, the energy landscape is evolving fast. Stay ahead of the curve with the latest strategies and solutions right here on the Hypercube podcast. See you next time.