IoT Leaders
IoT Leaders

Episode · 6 months ago

Maximising Asset Value: The Importance of Data & How to Use it Properly


Many companies are under-utilizing their data—a misstep caused by misunderstanding the value of the asset, how to properly mine the information, or how to blend the tacit knowledge w/ the explicit data.

Dr. Satyam Priyadarshy, Technology Fellow and Chief Data Scientist at Halliburton, joins the show to discuss his opinions and strategies regarding data and how companies can take advantage of the bulk of their assets.

What we talked about:

  • Making Sense & Insight Around Data
  • The Importance of Historical Data
  • Strategies for Mining Data within a Company
  • The Application of Artificial Intelligence
  • Hiring Different Types of People at Halliburton

To hear more interviews like this one, subscribe to IoT Leaders on Apple Podcasts, Spotify, or your preferred podcast platform.

You're listening to Iot leaders, apodcast from Si that shares real IOT stories from the field about digital transformation,swings and Mrs Lessons Learned in Innovation Strategies that work. In each episode,you'll hear our conversations with top digitization leaders on how iote is changing the worldfor the better. Let iot leaders be your guide to Iot digital transformation ininnovation. Let's get into the show. Welcome to the latest episode of Iotleaders, the podcast that aims to demistify the complex, intriguing world of iote. My Name's Nicole, your host, and I'm the CEO of Si GlobalIote Company, and today I'm delighted to have as my guest on Iot leadersDr Sachian Priya dashy and such a works for Halliburton and his title is achief data scientist and he's also a technology fellow at Halliburton. And if that'snot enough, he's involved in several startups and is a senior fellow at ajudge Mason University on Cyber Security, as well as a nut jung professor atGeorgetown University. So definitely a busy man and such, I am very,very welcome to iote leaders podcast. Thank you, nick. Thanks for theinvite. Certainly yes, or a little bit addition to the profile. I'mnot no longer with Georgetown, but I moved on and I'm now with Virginiaattacked, and Oklahoma State and university in India. So it's just like whathappened to in as a farmer. Academics you never leave academics, you see. Yeah, yeah, it's well, I don't know how you find timefor a little bit, but thank you again. I'm time for this halfan hour, so that we've probably been been together today. So such ait's a big subject and I never we spoke prior to this. It's apretty wide ranging subject as well, so it's trying to break it down intotwo pieces. Maybe I can just ask you have paps, just a littlebit about your own background before we be died in. How what was yourjourney like? How did you, how didn't you get to where you aretoday? Being cheap taking scientists for Haliburton and perhaps for those people who don'tknow Howiberton because most people do what they will be people who do maybe justa little bit overview of what Haliburton does in the ord gas bunch. Sofirst of all, let's talk about Haliburton, then I'll talk about my journey.Haliburton is hundred and two years old company. It's one of the world'slargest energy services company. I think one of the first pattern the company fildwas in areas of cementing by Mr Halliburton, and that's the company named after himand it is over the over the decades and a hundred years, ithas gone through a lot of expansion in different fields, but primarily it isa body call is company which actually collaborates, an engineer solutions to maximize the assetvalue. Now it's a very, very important part to remember. It'sabout maximizing the air set value. You know, in olden days we wouldjust call the hydrocarbon as an asset, but in today's world data becomes anotherasset. So that's just works in the ground in the efficient that the dayseris increasing to be instance, which is absolute. We're going to go right, absolutely so. And as at the company is Global. We have forabout Fiftyzero people around the world, pretty much representing most nationalities in the world. And of course the challenging task of brilliant completion exploration. All are allare very complex scientific and engineering based.

A lot of highly skilled people arein the company as well. So it's a great place in that sense becauseyou get to interact with mathematics engineer, scientist, physics geophysist. If youlook at the spectrum of talent that is there in the company, is significantand you call my own journey. So I did my PhD in quantum mechanicsapply to biophysics when I was pretty young and and pretty concept of how doif you think in a very simple terms, you know solar selves we make anefficiency of fifteen percent to twenty five percent, but nature makes solar cellswith almost hundred percent efficiency. So my PhD was like trying to understand whatis going on inside a chloroplast from a quantum mechanical point of view. ButI think I was way too ahead of the curve because it just can't computeanything, because it involved a computation of five thousand by a five thousand metricsand an and on. Sparse metrics you can compute easily because there are nosuper computers then an even today. Not Sparse metrics the challenge. And thenI switched on and I was trained by a advisor. I think I owemy success to him a lot because he trained me in such a way thatyou think of the problem not as single problem. It's a multiple problem issueand you should always be open to addressing multiple problems at the same time andas a result. In fact, during my PhD he said not to dowork on he won't give me these PhD unless I publish more than topic.You know. So, which which is really good because it opens up yourmind to different problems. And so I went and switched career and not carriers. I went to POSTDOC. Liked to postdocs in Australia in a totally differentfield, a glassy dynamics and lipid membranes, and then I came to us anddid work in DNA electron transfer and non linear optics and many other topics. So we have a very special model called Barton and pre Othershi model forDNA electron transfer, you know. But then I did a lot of supercomputing. But if you think our foundation is all in data, will beingquantum mechanist to generate a lot of data, and here I was the generator ofthe data as well as analyze, analyzing the data for sign very complexproblems. And so I switched my careers from there to become a technologist ata company called AOL, which is America Online. That brought Internet bombs andLucky I was lucky there that seven years I went from a individual contributor tobecoming head of research in two thousand and five two and set up a iceand perfectcellence in two thousand and five in Beijing and Bangalore, and then Idid my mba during that time and switch my careers again to become an executiveturnaround. Some companies got involved in startups and, as a result, oneday I got a call from Haliburton or a recruiter from while in gas industryset they're looking for someone like this and with its experience. I am havingfun for last seven years in Ali Burton, when I set up my center ofexcellence for a big data, data, signs and digital in twenty four,when nobody was thinking about it in the industry. Well, there's enoughmaterial. Leaver about twenty podcasts from flora plus and to Hollyburton and big data. So so let's let's just pick that last subject and go deep, becauseyou said they're instantly. He said the pink nobody was thinking about it whileI go in the subject the Biot, of course a lot of people say, well, I actually product for Myot is the data, although having itwould seem to be that in certainly in the field that you're in, there'salmost there's no shortage of data. They may well be a shortage insight asto what it actually means, but it...

...seems like the amount of data thatwe're nowt technically able to create it. I mean it's not a dripping tameor even running to tap it. It's not leave it the host pipe thatit appears to be almost like it's soun army, and it's only going toget more and more and more. I guess your work must rotate around thatin terms of how on earth can make sense and insight and interpret this datathat is coming out as in the context of the business value of the data. It's absolutely so. You know, people can challenge them on my commentthat people are not thinking about it, but that's not that is to anextent true. But then dustry have been general rating data, no doubt aboutit. Right and but if you look talk to experts, they will saythat or even today this is we're talking about twenty twenty one people will sayor data is of not good quality, data is not complete. So ifyou think of where the world is an if you'll look at data native companies, which Ewel was one of them, and now the googles of the world, you can call them now, if they also had the same problem ofdata not complete and a bad quality, then they would not be making money, right, because the money comes by analyzing the data, building recommendation,inn whatever it is. Now, if, in an engineering form, we keepsaying the data is of not good quality, then what have we doneto fix that right? And that's where I think will come back to morelittle details later. I think the IOTI is or the industrial iot will reallymake sense because that can actually improve the quality of the data. Right now, industry collected the data, and no doubt and the significant amount from explorationto drilling and completion and H S and other areas. But if you lookat it, the data was collected in the field. It was transmitted tothe back office at a certain frequency. So not all the data was evermoved back office, but I if it was moved, it was moved ina later stage, so to say. So there was no real time soto say, analysis of it to the extent that it could have been done. Now, sir, if you look at even the last twenty years evolutionof the technology, for example, such technology like we all depend on itright, for all our practical life. We depend on it. But whenit comes to implementation in the energy sector, that has been really poored. Ifyou go to any of the big energy companies and you want to searchfor him, say, a rotatory pump that you have used for last tenyears and you have repaired it twenty times, and if you want to search forthat repair log you will not find it. Why? Why? Wehear these stories a lot about it's almost a paradox in a way that thecompanies that arguably could benefit the most from the data, like this device hasbroken twenty two times on the run. It's not broken for the twenty firsttime. Can I just find out what to do? And that would sayme a lot of money. And there's so much money involved, and oilexpiation, for example, is a massively capital intensive, it's very expensive business. It's all about classing. The time you hear these stories and then youalso hear not just what a better but large companies in general often are notthe ones who have captured the learnings, captured the knowledge, and they arejust some extent behind. Why is that? Is it a cultural thing? Isit? Is it do you believe? Is it a volume of data thing? Is it that they were just focusing on something else as a priority, like finding the the Oily Gash? What do you think that is?So I think my might take on that is that, you know, whilethis technology evolved, the industry did not spend any in like did not spendwell on actually understanding these technology. Well, right, so, whoever, avender came and said, Oh,... is my search platform, youcan implement it to search your corporate website. Yeah, and that's of God implemented. Nobody looked at the use case, how it could be really will bedone for other areas or on top of a database. Right, sothose two are not thought about very well. That how how the value will comeand also one of the culture. It's a kind of a cultural thingand call it a business strategy. When you are in an operational mode,your goal is to fix the problem and move on. Yes, but ifyou look at the whole framework of data science, is that you want todo? The science on the data. That means you, your historical datais your really good source to do that science experiment right, because you alreadybroke your pomp, already broke Ye. You have been of the data.Now I can actually analyze that data and understand that for last five times,why did it break, so that I can avoid those situations or those conditionsor can address those conditions well in Ad Mans before it happens the next time. So the thought process has to be there. That historical data has significantvalue and that value could be measured. And in fact, when we didsome experiments, I would call them experiments, in early two thousand and fourteen,we showed that how much valued is there in terms of monetary value?That the that the organization saw their come. Industry has lost. So the sittingsaid, yeah, sorry, they're sitting on an ass at the hugeassay which is historical days a but in general people are looking in the directionfor the value developing in the other direction, all the new things the channey needsor what they're doing next, new implications, but actually lot of valuein the legacy. Days it's just sitting run split and I'm sure in yourinteraction you must have come to people asking the question who owns the data?Yes, right now, if you think of as a business and you're beinga CEO. So if in principle, the company owns that data, yeah, right. So why is this question about who owns the data within acompany? So when you start thinking that the data is owned by the company, then you start analyzing it and that means people should start collaborating. Infact, I wrote an article two thousand and fifteen, I think, calleddata democratization, and initially there was a pushback thing. We can't talk aboutthese things. But democratization doesn't mean to give away. And how can welook at that? Need that idea. And and you're in a workflow.You have US three steps Moret of data, somebody has another to next three stepswork of data. When you look at the whole workflow holistically, thenonly you can create dvial when you're doing things in a silo, you reallycan't create value too much right. It's a finely. Excuse me, sucha help. I mean this value I think knowlergy have got my mind,is that was gold under our feet. We just can't get hold of it. But there there's value in the data. This huge amounts of data. It'sowned by the company, but but there it's not like there's any issuesin accessing the there. It's already their data. So where do you start? It seems like what you're addressing, your role in Halliburton, is sovery big problem. But it put a very important problem in the very valuableproblem to solve. But it's also a very broad problem. I mean there'sdata everywhere. It's it's everywhere and there's Timor creating it every increasing values ofwhat. What advice would you give on the people who past listening to this? Think you will. Yeah, Michem is in the same situation and wegot all this data, we got all this new data. Without getting allthis needs technology and sensus. How do you go about getting strategy to mindthe value? Yeah, so you see what I have been doing in forlast seven years. I'll say from that...

...practice and I've been done this forother industry before, so I know that it works. You see, useif one is you have as a CEO or as a people be under CEEO. People already know where the serious problems are at. It Act at atacit knowledge. They may not have a quantification of it precisely because it's agood idea exactly. Otherwise they wouldn't be in that seat. Yeah, soyou look at the experts saying, okay, in the last five years, sixyears, what thing could have been really improved? And then you breakdown the problem into what it called the sprints, as you call your andyou want to run that marathon, but you want to break down in sprintsand saying, okay, I will take let's assume, take this example thatI started with a fair rotatory pump. Right, if you are the ifyou have deployed that pump for twenty times and it has broken ar say ata at a rough asset knowledge perspective, it has broken after every three monthsof deployment. Yeah, you have some idea that it breaks after three months. Now I want to really narrow down that problem. Why does it happen? So you come up with them, you come up with a solution ina way that I can look at the data close to the three month period, five days before, at ten days before, a twenty days before,and see what happened really in that operation. So you basically what you do isyou take a business problem which is you can say I will say fiftymillion dollars, let's say. And then you say you break it down theproblem in two chunks and saying I'm going to look at a problem which willwhich I can take a data worse six months and if I can generate twopercent of that savings, then it's a problem worth solving. So the conceptthat I say is that look at the data from a proof of value product, not proof of concept concept. is well known that data works, thedata science works right, and the data that you have actually, how muchvalue does it have? And then you start integrating data from many places,saying, for example, if it is a pump, you can connect whetherdata, you can connect, if possible, hr data, right, or connectyour chemicals data, you can connect your deployment, repair, log parts, log whatever it is. you start connecting different data sets, saying orwhenever we replace a bearing of this model, it fails. So you have tostart asking questions from the data and then adding on more data sets.So you do think this craft best of fashion? Very yeah, so firstphase should be less than six lessons, around sixteen weeks or less, ona very small amount of data where you show, yes, the data hassome value. Then you add more data than you do another small project calledproof of value. Then you say I need these five more data sets toconnect to it. They could be under different silos, and then you scalethe problem so by the third step you already know how much value are goingto generate, either in cost savings or revenue generation or accuracy or efficiency orMPV. So it is a anything you do with data has always value.So were you know, you may have hard people saying on my digital projectfailed, on my data sience project fail. I don't believe in that at allbecause no projects fail because every project has a value. You could notscale it. that a different issue. Well, actually, some of thedata that my company, a side that I remember and every bat companis actuallythe day to the data on the failure rate priority projects is appalling. It'scivilians real is simplorked. About eighty percent of IT projects never make it pastthe Pfc. when you do a click on that, the prefect concept youwhen you double click on that. It's actually not really a technical problem.There are there are issues definitely to do...

...with the device. Most people don'tknow anything about how to design and don't want to know anything about how todesign, and that's a gap that we that we filt working with the moduleand Eventualis, for instance, the quick terms of the world, which amountthere's the world. But also it fails because they certainly it's like to anyour tap on. They suddenly start collecting a lot of data and that's thepoint at which they just breeze. I mean they just they can't measure thequality of what they're getting. They don't know which date is important, whichthey is not important. They haven't got an architecture for what are they processat the edge? What? What data do they back or if they sendit all back to head off is partickular. Haven't imagine viewing an oil field andrig or something, the amount of data and Terri Bytes, paeda bytesand we are the amount. You just can't afford to send it back toyour corporate headquarters to crunch it. So you've got to do edge processing andthey just they just didn't think about problems of a data architecture and big dataturn insight when they started there their project. And so you do see what you'resaying in the General Statistics on the industry, in that people they don'tstart off with trying to drill it down to let's just try and find outone particular problem with good chase down that will particular problem. Often they startoff with a horizontal approach and say, let's collect data, promise many thingsas possible and then we'll work out what to do with our data. Andthat's where often they just breathed and they think, I'm not I don't knowwhat to do. Is I'm collecting data, but I don't know, I don'tknow what I'm going to do. Yeah, absolutely, you know,but my philosophy has been very different. You know, we can address asbig as of a problem they in today's Today's environment. When I iote sensorsor any device that is generating the data, data can be in any format today'sworld. Come because compute is so cheap, right and and in principle, you could really put out very what I should call powerful machines at theedge as well, in a very small factor, so you don't have toreally back hall everything. Yeah, right, whether it's cloud or edge or whateveryou want to call it, found known and you can do that right. You can develop the algorithms or the models, whether it's scientific or augmentedmodels and you can push them back to the field, where where the challengecomes is that we really can't make automated yet because we have to really testanything we do in the field and any complex industry you really have to testand validate. More often than anything, it's right. The models are modelsafter all, and more validation is done, but the size of the data isnot should not be a concern, because if that was a concerned then, as I said, it and native companies will not exist because the problemhas been solidage. Is it a rating Fast Enos? It's it's not inhere. Its just because the emagine, especially with clouds, the amage ofprocessing this available is not a it's not a blockage in the process. Iswhat you're saying. A salute here. It's more about business problem that wewant to solve and whether I need to really look at the data from onemonth or should I just only look at a one week, because depending onthe worklow we are talking we could be really creating more value within one week'speriod of work of data and should be good enough for us. Light orsomething. We might have to look beyond it. So most of the thingswe want to eventually do in real time. And so that means you don't don'treally have to process Betabytes of data all the time, because no,we are not. We are not. Any of these industry may be,say refinedies. If they were to deploy Fortyzero censers, maybe they will geta daytime terabytes. Because no, ioti sensors sense data in Giga Bytes.Yes, per day, I guess it.

Just think that they another hour tothat. This is probably speaking to the personal does an I will ring. I heard and it was a perios and oil ring all it sees me. No, refine or could have ten million census in it. Now,after a feeling, that's a better four five euros. Just say, isthat the air or to spokes to pay for you say so. I've beenyou know, these are numbers written by a lot of people. I'm notsure. Sure, I've never seen them before, like I've not been toa field, so I can't say that for sure. But what how dothey calculate? And what does the device means? What does the sensor mean? Are they really internet lot of things devices, or there any devices?Right? So it by definition, anything that is connected becomes an Internet ofa thing or Indian right but if you are, if you have pumps thatare running and generating data which is collected by hand, that that's not Internetof a thing, but it is a data yeah, right, so theseare when people write these kind of articles. I don't know how many of themare really counted. What is connected and what is not connected. Almostcertainly nobody be worried the IOG business and we have customusy. We or guess, and I let me tell you, in terms of true IOT devices asopposed to something with a control and it's able to spit out data, whichisn't Iret, but it says true I devices, it's probably the hundreds inpractical terms today. I mean it's nowhere. There said it might be in thefuture with maybe with big when we start getting product for gene networks inthese locutations. But Eve and then I think the word phrase Iot is beingstretched to cover everything at Troy. That's not what we're talking about. Infact, is counted productive to think that that's what we're talking about. Absolutelyand I think again it's a future of the web, when everything becomes digital. Maybe that is when we will get that kind of censors account of sensorsand connectivity. But we are. No, we are close to that right inif you look at land as industry, they're talking about digital oil feels twentyfive years ago or something and integrated reservoir management some thirty years ago.So what is integrated and what is did which digital oil field really exist?Right? Because Foundation when you say digital, I I feel if everything is connectedand you are really you're doing real time automation, that's than it becomesreally valid. But pieces of the puzzle are automated, no doubt about it, but we don't have it fully holistic automated digital oil fields. Well,the other thing people set it up that allows transitions to one of the finalbig subjects was getting you eat it about crazy is got twenty years ago.But let's just say by days good people would say. Well, you know, the answer to this is nothing to do with human as you were saying. People would say by now or in the future shortly, it's all goingto be artificial intelligence. It's going to be machine. Machines will say go, the humans will let go, they'll stand back. We were worrying aboutwhat we're going to do for jobs, because it's all going to be brother. You know, the machines are going to beat the humans at the analysis. They can learn about the Poms breaking, they can learn about the resolution,they can go they can give you from reactive to proactive, preemptive.Mean that's what we they see in people's cause they see it. It willbreak, and so you take in Stemhech we fix and then it would be. That would be reactive and productively. Ali Go on saying it's going tobreak, that it's going to oil. So put oil in before breaks andpre empting the test where you get any cardoon get. It's like the icon. While you were sleep, software update downloaded. We fixed a whole bunchof issues you never even knew you had.

But don't worry, you'll never havethem anyway. Have a nice day. So by now we were going tobe in this world where, or at least entering into this world,the machines and take over. No, I know we spoke previously you andI can see it is one of the I never we spoke you. Youhave a you have your doubts where not of the world of Ai, butyou INSETT me. You didn't even like the phrase ai because of the yourexperience or the expand on that little bit. Yeah. So, so artificial intelligenceby itself as a subject has been there for fifty plus years, right, and if we look at even the applications of Algorithms that are developed,it has been used by oil and gas industry for last forty five years,whether it's neural network, whether it is regression. All right, it doesn'tmatter which algorithm you're talking about. The world change on the technology side andthe computer side. An artificial intelligence is just a subject, right. It'slike my analog G that I've always explain. We never say that we are eatingchemistry or we are wearing chemistry. Right, we are all closed.Are made out of chemicals. Food is made out of chemicals. Some chemistrygoing on. Right, application of chemistry that we're talking about. So thisis in the same way. It's an application of artificial intelligence, whether it'srelated to audio, whether it's related to video, whether it's related to data, whether it's related to text, text, right, that is what you're talkingso there is nothing, there is no box called artificial intelligence. Pain. And when these articles come out saying artificial intelligence in test law, isthat the same box? Can I put it on my computer for a fardoing my statistical analysis. No, right, so it is not a thing andthat's where the confusion is. But in aspective of that occreates a fieldand it's important field and it allows you to analyze things that human beings bythemselves could not do right at scale. Repeated task that can be optimized,and even can eventually when self learning algorithms will be there more matured. Thenmaybe things can actually improve, like you can see those examples in robotics alittle bit, whether robot can learn and things like that. But where arewe? We are far away from it in terms of application. Maybe,maybe in some defense sector things that we don't know. Who knows what goingon there, but in a practical in a practical world, that's no.We're close to it because it requires, it requires in principle all the tacitknowledge that is sitting in your head and all the people who are actually inthe field right what to do when that is in people's and engineers had,after two thousand and thirty, forty, fifty years of experience, and it'sthe stupid and if we look at the history of knowledge management in the companies, we have never figured out a way to capture the tacit knowledge. Knowledge, yeah, and that doesn't knowledge is what is really what algorithms will needto really make a decision. So back to your example about the pumps,weters. Well, asking the engineer has been around bridge is stability. IfI was going to investigate the plate on one area, what we today experience, engineer will say, you really want to have a look at these pomsbecause they break three three months and it's a really big issue. That's itcomes from the just side. Experience. That does in knowledge. The chancesof the computer system saying that to you are probably pretty pretty low. Andand then when they do present with the data, they need that tastit knowledge, that experience, the thing that we can't codify to actually interpret the dataand prosise the actions. And so it's a company, you're saying, youguess it's a combination of the two. Absolutely very important combination, especially inthe island gas industry. Having worked in like you know, and consulted inalmost some and adverticals before I came here,...

...and I can tell you in Alanhas industy the people have so much knowledge of because the processes are complex, irrespective of what section of the work work life sorry, work flow.We talk about in general the energy sector. If you look at it, it'sa very verd you call science and engineering driven industry, and so alot of these people have so much tacit knowledge in them that really needs tobe captured and can be taken advantage of. For example, when you're drilling right, if you think of it, the person can feel and say Ineed to rotate this much the building. Yes, now to to get aalgorithm to do that, you have to really look at so many things.First year to understand what is the force coming back easy, and then youare to really analyze what I did in the past. Few scenarios like that. So it will take some time and and that's where I'm saying that thisyou'd but at the on the other hand, those people who are experts, theyhave this sense that they feel this, see the sound or the field ofvibration and they said do this now. They don't believe. The data,the data of people don't believe don't have the same knowledge as the engineer. So they have to come together and I feel that the people who havethat tacit knowledge, they can be trained with the data knowledge, and thatis what I call the talent transformation process, and I a sally, not theother way around, the other that round is hard because you you can'tget the field experience and it. Yes, yeah, you can only get whatthe feeld people tell you. Yeah, but that, but the few peoplehave so much, so you really need to throw them a laugh thing. This is what happening here. When do I do? And then youfortify that it was last question that, because it's opens, is as up. So may different questions. The combination of the field experience to give youthe tested knowledge and the ability to make that instinctive human judgment that says thisis right or this is wrong. The vibration. It feels right, doesn'tfeel right. We don't know where it comes from, but we can doit. And then the machine playing its role, analyzing things very, veryquickly as well. Does that mean that in your role at Halliburton, doyou train people? Do you recruit different types to people for the world thatwe're heading into? You know, do you look for certain types of degrees? And reminded listening to do not in the field of owing gas, butyou know, conversation. Certainly, I've been involved in for many, manyyears lots of different industries that I've worked in, people saying, you know, NBA is NBA students are useless because they don't have any of the practicalexperience, but that they whereas cos the NBA schools, will think that they'rethe train of the future leaders and everyone's got all the knowledge they've ever didbecause they got an MBA. You know, you take that you've never applied itto an oil engineer. So and often, like in the case aboutthose or those who went to work for several years when did an MBA andactually she felt at least she was much more valuable at the end of thatthat if she's done it the other way around. So do you in howyou can get make recommendations in terms of what types of people you employ,given in this world that were in already and heading more into, or doyou actually run internal training courses on have a hit this combination of the machineand the Tusit and the human working in harmony? Yeah, found answer allthose parts in a interestingly so as an academic professor, all my students Ifor MBA especially. I tell them there's no point in doing MBA after bachelor'sget right couple of years of experience. Then do an MBA. Then youwill know what mistakes I did or what you didn't do right. She gotit, and that then the value of MBA becomes really important. Otherwise it'slike taking any other courts. You passed it and you're done right. Andthat's my first recommendation to most people in...

...terms of since in my background wasnot a direct vile and gas, so I know that it's all about generatingvalue from the data and I think in that way. Most of the teamthat I built initially are all people from different fields of science or engineering orother areas. So I have a PhD in atomic physics, Slash Aster physics. I have a PhD in chemical engineering, mathematics, economics, things like that. So they can look, they can think totally differently, but thenyou pare them up with the task, a knowledge people that subject matter experts. That helps, and then over the years we actually developed our own trainingprogram not only for individual contributors but all the way to the leadership, becauseone is you have to really keep these people also in house. Right there'sinteresting challenges in the world and if especially a data science people, their high, high end demand really. It is our jobs out there offering future rangesexactly. So the way my philosophy has been that for all the data scientists, give them interesting problems. Don't give them and put them in a boxand do just one problem. If they are doing multiple problems at the sametime, there's no problem with that. In fact they lock because then theycan think of it over that I have this kind of data or these otheralgorithms working, but I have this kind of data wise is not working.So they have their own compare and contrast going on within themselves and and thenthey're interacting with different domain experts, so to say, and that helps themreally think beyond a simple problem and then they it's an exciting environmental work andthat's how we have grown this center in Bangalore and in Ustern and Columbia andmany other places. We are working with so many people. But the trainingpart we've developed is because the same people who are actually working on a problem, they are actually teaching the House and House of this field to the domainexperts. So they when they ask questions, they learn from the domain expert.Why are they asking this question? Why can't I find this and whenthe domain experts is how to do mathematically or why it is like this.They can explain it and that this synergy is significant. And last I thinklarge. Just in last two years I think we've trained over thousand people inthe industry and so and and hence I don't really worry about the talent poolside of it. In fact, I'm one of the hats I we areis the managing director of India Center and in the last one year I hired, of you, about hundred people from all different fields. So so youknow, it's a fascinating area to work in and I think the potential issignificant. As I say, the opportunities are significant because we are only scratchthe surface of the of the industry. And if we really have the desireto build full implementation of IOTI sensors, properly, getting the fives network workingor beyond five g working, which will reduce the cost to move the dataand bring the speed to the connectivey to then I think we will have tobuild what is called a digital twin, of digital twins. And so there'sa fascinating field and of course that T's a knowledge is not going anywhere.I do I call it augmented augmented analysis going on. Yeah, it's what. It's a fascinating story. It's a fascinating journey and it's also reflective,of some way of of Halliburton's journey as a company into moving more into dataand data services for the clients and, as you say, all the efficiencies. And then the whole subject of digital twins is something that we do plantogether as well in a future podcast book. For the moment, we better leaveit there because we do it so...

...much, so much grand. So, Manya, I can just finished bride the subject, thanking you for yourtime and sharing with all listeners your journey, what you're doing in your thoughts onhow to go about it, and for also astrailing those people who arehave we're out there being concerned about whether the machines will take over the actually, you don't believe that they will and that we're all going to have plentyof do going forward in future years. So with that, I just wantto say thanks to everyone for listening. You've listening to the IOT leaders podcastwith me Your Personal Gil. If you have any feedback or questions on it, to remember that we do have an email address which is iot leaders ats I think that's e Se Yecom so we loved to hear from you andhe suggests from many subjects that you would like us to cover. As youhave this this is about guest we can actually go very broad or reading verticalinto industry and we love to hear from you as to what you would liketo have a discussion about or even when you feel you'd like to be aguest on the show. So let's beat it there. Such a thank youvery much for your time and for our listeners. I will see you andtalk to you on the next as. Thanks very thank you, nick mynot. Thank you. Thanks for tuning in to Iot leaders, a podcastbrought to you by SI. Our team delivers innovative Global Iot cellular connectivity solutionsthat just work, helping our customers deploy differentiated experiences and disrupt their markets.Learn more at SICOM. You've been listening to iote leaders, featuring digitization leadershipon the front lines of Iot. Our Vision for this podcast is to beyour guide to Iot and digital disruption, helping you to plot the right routeto success. We hope today's lessons, stories, strategies and insights have changedyour vision of Iot. Let us know how we're doing by subscribing, rating, reviewing and recommending us. Thanks for listening. Until next time,.

In-Stream Audio Search


Search across all episodes within this podcast

Episodes (20)