Beyond enterprise data platforms: Drive success in data-to-decisions journey

[Music] thank you my name is pankajuk I need the Google Cloud business unit for LTI mindtree today we're going to talk about a very interesting topic very close to many of our hearts uh you know many of us uh have in the recent past setup Enterprise data warehouses Enterprise data platform a lot of Empirical research has been done in the past few years on how important data strategy is to your business strategy and how it actually creates true differentiation in the markets um and many of you who and many of our clients have actually set up these Enterprise data warehouse platforms and and they spent a lot of energy doing that today what we want to do is we want to actually expand the aperture of your strategy in two different ways and we talk through that a little bit and we will actually you know with me I have uh actually he leads the the data and analytics practice for LTM entry and we also have a niche Anish actually is the CIO and CDO for genius bank and he has a very very interesting story to tell us which actually kind of will cement the thoughts we're going to talk about so these are the two these are the three of us uh you know my picture is five years old so I think you should be able to figure that out um actually maybe 10 years old so so that's us so so now let's talk about the and we'll talk about the two two ways to expand their per so let's just start over the first thing so this is uh as you can see in the picture this is actually Mount Everest so there was a research done by the British medical journal about six years back of all the fatalities that happened from 1921 to 2006 on Mount Everest 192 people died on Mount Everest between 1921 and 2006. what was most interesting was that 56 percent of those people died after they had ascended the peak of the Mount Everest which is 29 000 feet they got to the 29 000 feet they celebrated and I think they kind of forgot that they had to get down or they actually didn't plan for getting down uh if you know a little bit about Mount Everest or any other 8 000 8 000 feet mountains uh the death zone starts at 26 000 feet so as people were starting to get down 56 percent of the people actually who died actually died died when they were getting down so it's really really important to realize that once you set up your Enterprise dataway house there's a lot of stuff to be done so you need to celebrate of course but you have to think of the life after very very important so that's the first aperture we're going to actually talk about so think ahead once you do that once you set up your Enterprise data warehouse platform think ahead think of the life after so that's your first aperture of expansion of the aperture of thinking when you when you think president Warehouse strategy so that's number one now I'll talk about the second thing as you guys see this beautiful animal is a dragonfly there was there's a lot of research being done on the hunting success rate of animals so if anybody wants to take a guess how much is a the hunting success rate of a tiger I'll give the answer it's actually 15 percent how much is a hunting success rate of a lion or a pack of lions lions or lionesses hunt in packs it's 25 percent peregrine falcon falcon which actually is a very very bright and the fastest one of the fastest birds in the world they actually dive down they actually dive down and then they actually they go for the kill and they actually shoot back up again they are hunting success rate is 47 percent you look at all the animals in the animal kingdom the hunting success rate will range from 15 percent to to the top of the pack is the is the wild dogs it's 67 percent there's one animal which actually stands out and that is dragonfly the hunting success rate of a dragonfly is 95 percent now I'm why am I telling this to you I'll tell it to you in a second but you know but it's just very very intriguing so actually Harvard medic Howard uh Howard University actually did some research on um on dragonflies and they try to figure out you know what's going on you know how are they so successful so what really happens is that a dragonfly has a compound eye it has 30 000 facets of where the information comes into the brain and it covers almost 360 degree of of the environment around it the information comes into the brain it gets processed and it has four wings and those wings work very independent of each other so what's really happening is the animal is wired to be able to take information and that information actually is 360 degree information from a whole variety of different sources it gets processed and the wings work independently and therefore the dragonfly can actually power up down left right and it can just stay around so what's really happening is your information is actually in many ways getting captured from a variety of different sources process and and executed on in a very very differentiated way now the reason this is important is when we think of uh when we think of our data strategy we think of setting up the Enterprise data warehouse platform what we forget is the act of Thinking Beyond which is actually think the entire Journey end to end think Beyond what information are you going to get how are you going to process that information and how will you execute all that information so that's the second point I wanted to make so when you are thinking of your Enterprise data warehouse strategy there are two things you need to think about think ahead what's a life after once you set up your Enterprise data warehouse how are you going to actually manage your Enterprise data warehouse after you set it up and you have to think before you get to your Mount Everest and celebrate yeah you should have thought about it do you have enough oxygen have you thought through that or how are you going to get down you know are you actually going to get down on time how will you actually work in a team or whatever else you need to think about but think ahead and the second thing is think holistically think every aspect of your entire Enterprise data warehouse including what information are you going to seek how are you going to process that information and how are you going to execute all that information now to actually cement this I'm going to actually invite jet is actually going to talk about uh how you execute to these two expansions of aperture and then after that I'll invite Anish who will actually talk through a specific very very interesting story which we will talk about how we actually do that so just over to you okay uh thank you so much for taking time in the morning I know yesterday you all had a very interesting Keynote and the reason you must be wondering some of you about starting with Everest to dragonfly what's going on here I think the idea is about we talk about in the data World narratives and storytelling and bringing to the attention of people why you need to do the right investment I hope you are done really reasonably good job in terms of getting your attention about why this session is started with some kind of a story uh I go with my name Jeff and I've been in the industry for last 26 years doing data analytics and I for many clients over the years and I can't tell you this is the best time for people who are in the data professionals who are investing in data and trying to drive for the first time everyone is talking about how do I really get best out of my data and how do I drive more monetization and I'm sure every year there is a new buzzword one year and it's a return is the new oil another year data monetization data Marketplace data products I think you name it we have gone through all of that I think this year probably the the new story is going to be Janai and I'm sure each of you has seen very interesting demosis today and some of us got a chance preview of that much ahead of time we were able to bring that in our Solutions and started driving to the clients so what I'm just taking the cue from what Punk had started about all of us know in the industry data Investments been happening over the years and there is nothing new what we are talking about but what is new and what is old is I'm going to quickly touch upon today so if oops sorry yeah sorry so if you think about there are a few things which all of us know over the years and I think it's always important to talk about those basic I think my friend from my good friend always talk about three things important in life which is about having a good slave good food good exercise all of us know but uh many of us don't do all the three regularly the same way some of these points all of you know it's not about you don't know or we don't understand but over the years there is not enough time spent on these areas by each of us and our investments so let me touch upon each of them in a very clear manner the first one is about user adoption if you heard about any horror stories about data implementations I think one of the big reason is this the user adoption people are used to use a particular technology people who are getting information from a particular area suddenly we go and talk to them about we are migrating to cloud and once it is done everything is taken care and they don't have enough Comfort once they go to the new area whether it is because of lack of understanding lack of awareness whatever is the reason so one of the most important is a user adoption and evolution as the users start using I'm sure you all agree with me they become more hungry about the data they would like to get information much faster so part of whenever we think about new implementation and new modernization it is very important for us to put enough effort on that number two just getting into an attention getting synops right and I'm sure those of you heard about the since implementation of cloud our costs are gone up any of you face that and I'm sure quite a few of the customers went through that kind of a journey the reason for that is maybe it is not well understood well thought through not an architecture is done for the right reason maybe simply a lift and shift kind of thing is done and also there is some kind of a misnomer myth people think about going to cloud is going to give infinite scale so people can play whatever they want people can explore whatever they want and may not require all of that so it's very important part of this data modernization we need to set the phenops right so otherwise people tend to give Cloud implementation or is not successful it's nothing to do with that cloud it's to do with the way the setup and architecture principles has to be right so that's the second you know I wanted to bring it to your attention I'm going to cover a little more deeper not living at that headline but I just thought that's an important aspect the third one is about the data quality and observability whatever name you talk about the data is getting flooded in each of our organizations and I'm sure uh you all seen the same updates about how many Zeta mobiles better miles that's happening even in the urban organization day to day so it's important to have as we invest on these areas how we are able to Leverage The end-to-end automation of that information if I have to give those of you watch Keynotes today if your entire operations you're able to get an audit you're able to get end-to-end view you are able to get solving those problems in real time how good it is and that's an area you have to think through and each of us wanted our investments to get zero Ops no Ops or very less Investments on the Operational Support so that's where data quality and data Ops is going to be very very critical and the only way we are going to get users Comfort is about quality of the data and then finally these days more important is about security and privacy all of us seen n number of incidents and the regulations and others so in a simple way I will talk about the usability and then the performance which include the cost and then the quality and then finally the privacy these are the four big things all of us uh you know have to really think through enough while going through this journey and again as I started with it's not about we don't know but often they were underestimated when we start our journey and most often when you go back and root cause analysis of a programs which were in difficult situations by any organization you may end up finding one of them or you may find the implementations are done well but over the course of time you have written on investment is not really what you plan for and again it boils down to one of these things why are we talking about all these things so if you think about user adoption um those of you have any of you I'm sure I'm assuming all of you have completed some form of data modernization data migration are setting up your data to drive why do I do that why are we investing so much on going to Cloud why is there so crazy about everyone I think bottom line is in my view two parts one is about data democratization going to Cloud will help you if you are a large conglomerate you're an MNC or you're a company which is part of the largest supply chain you can exchange information you can share information so data monetization is one big thing and then naturally the next big thing is going to be how do you really make sure you are able to give the information right time right place at right level of you know light level to the people so when we go to modernization one of the thing which we found quite a few times is about you have the users who are using the reports and there are quite a few people who are doing Downstream systems where you are sharing information and there are set of people who are your data scientists and ml Ops engineers and others so you have a different types of personas in your business and often times we realize your modernization you are not taking care all of them in terms of what you need to do so it is important for us to look at what are the different personas and how each of them are going to consume that information and it's not done properly we found many times organizations are going through suddenly they realize we have some Downstream systems they're still looking for FTP and it may sound very weird when people are talking about Cloud Journey but unfortunately world still has a lot of such systems now you cannot postpone your production go live out of the blue saying about your Downstream system so waiting for it so just to get your attention it is important to have the right personas it is important to have you know understanding with each of them I'm sure all of you would have seen the nudging the users and bring making sure users are used you know used to a particular screen a particular technology for last 10 15 whatever years they've been using their reports for 10 years 15 years suddenly we go and talk to them about start using Lookers start using something else it's going to create beautiful features while it may be great for us as a technologist or as a business people who are implementing the programs the end user many times feels it's not as easy as it is for them either two reasons one probably they're not been involved in the beginning or second they don't have enough comfort with usage of it so the folks one of the things which I wanted to bring is about we need to plan for that and we need to nudge and we need to ensure some Champions about it the next one I wanted to bring it to your attention is about the phenobs I think if there is one area which we have seen significant amount of interest in last one year and this is one while the user personas and others are always there but synopsis The phenomena as more workloads are going to Cloud the Investments are doubling tripling for the organization and what does that mean and I'm sure some of you would have read IDC and bunch of other analysts talking about 40 percent of cloud Investments the organizations are going through difficult simply because a huge amount of burn is happening then what they plan and when you plan your return on investment that's not what has been done so it is important to have your phenops ahead of time and have a clear structured mechanism while you give Comfort to your end users Comfort your line of business how they can use the information but it is very important you have a monitoring of that and I think we need to help and those of you tried and tested suddenly a Big Bill is coming with your credit card using a one of the cloud I'm sure you all of you gone through or all of you seen in your history and I think that's where I again insist each of you to think about phenops is the area where your implementation can be not just successful but it will sustain for much longer term the next thing I wanted to bring it attention is about the data quality and did a governance piece of it I think often they have seen over the years the data quality if there is one thing where you get a trust with your end users is about data quality and these days we are no longer using Data Warehouse just as a warehouse whenever we have a time we will look at it today we are using data warehouses data Lakes as a mission critical some of the organizations looking that for their financial statements some of the organization looking for real regulatory some of them looking real-time supply chain efficiencies some of them looking at real-time credit card fraud kind of thing so this data Estates are mission critical so in a mission critical environment you cannot be you cannot have a situation where your data quality it's not a complete it's not accurate so that's where I think enough importance has to be given not when you are doing modernization but once you complete modernization you may go through your organization may go through mergers and acquisition your automation may go through new implementation your organization may go through a set of new regulatory requirements so it is very important to have enough Focus right up front and I'm sure all of us know there are enough methodologies there are enough Technologies and toolkits available even within the gcp ecosystem so I think a lot of investment had to be done there again folks this is very very underestimated during the implementation by every organization and whether suddenly we keep hearing about model drift data drift and you know there are some challenges in our system our cxo is still very unhappy with what is happening and we are still trying to find out a single view of our customer we are still trying to fuel single view of our products all those challenges during modernization if you're not taking enough importance you may end up with that so that's where it is important to have data quality and data governance uh in area and I'm very happy to say today luckily a lot of AI and gen AI is really helping us what I mean by that is we we used to many many years back those of you started more than 10 years back 15 years back 20 years back we had to literally manually capture the metadata manually capture data cataloging and today we can use Chennai to scan to give our code scan your databases and actually help you to drive your metadata and data quality and data catalog much better way and that's an area when we talk about data quality we should use the new tools and techniques to solve the old problem data quality problem always existed but now we have new ways to solve that much faster but I think that should be part of your uh you know ecosystem and then finally everyone wants now a data Marketplace and I'm sure all of you would have heard about how we can actually make more money by monetizing our information now if you think about this is where it's very important for a data security and privacy every organization wants to make money from their and I'm sure each of you would have heard from your cxo saying our competitor is making much more meaningful out of the data they are using the data to build new products they're using the data to improve their operation they're using the data to actually drive more Revenue I'm sure all of you have heard about and these days all of you have heard about data marketplace where you actually can share your information in a Marketplace and get some users consume it and get the data out of it the same way you may be consuming somebody else in the marketplace the information so how that will connect the dots field so again which has been very very fascinating for everyone to make money from this but very little understood by most of the organizations how do you begin this journey am I going to share information which is going to be competitive you know then am I going to lose my competitive Advantage do I keep all the data in the marketplaces how do I get the security and privacy of the information in this whole journey I think we keep hearing about if your organization is a multinational working in 30 countries 80 countries 100 countries these days it's become a very very easy for every country to come out with their own data privacy and their own policies and it's going to be very difficult and each of you in this room may play A Whiter role in terms of how do you really bring that so that's why the data privacy and security plays a major role in helping your organizations to not just monetize the data but make sure you don't get into any regulatory hassles you're going to get into that so in short folks uh what I could impress you is about the four pillars which you talked about the usability make sure your new Investments are usable otherwise you just dump something from somewhere to somewhere it's not going to bring number two make sure your Cloud Investments on synops are super plant ahead of time number three data quality is the only way with which your Investments are going to be sustained and then finally it is important to have your privacy and Security in place with the amount of challenges world is seeing now and also genii is creating lots of opportunities for all of us in our business but at the same time there is lot of guardrails are required and that's why it is important to plan ahead of time and do this so that your entire Investments on cloud will actually exponentially give your written on investment so that's some of the things which I thought give a perspective you and then final point all of you would have heard about decision support would you mind raising your hand decision support okay now from decision support now we are talking about decisions augmentation and then finally we are talking about decisions automation we want to reach a stage where with AI and gen AI your data estate can start become a decisions automation where your company will start driving a business transformation truly not just a lip service or business transformation that's what I thought I'll give some perspectives very happy to hear some questions later part of time but I'll pass the mic to Anish thank you thank you good morning thank you pankaj thank you chit for setting the stage um so now let me add to the story with uh genius how a genius we are looking at data driven decisions before getting there let me give an introduction I am Anish Jacob I lead technology at genius Bank and I will also give an introduction what is genius bank before that I have to talk about smpc which is our parent so SMBC is one of the largest banks in the world 12th largest bank by asset operates over 40 plus countries in the U.S it's been operating for more than 100 years primary on the commercial business genius bank is the retail division of SMBC in U.S we started building this two years back we launched our first product and secured loan two months back next product is ready to go which is savings and we have like full spectrum of full banking product for next 10 years so it's sometimes very rare opportunity to build something from the scratch when we got an opportunity to build something from the scratch we want to do it little differently we want to make better product better experience better rate better data Insight that users can use and make some actionable actions based on Insight insights they receive so when we look at these goals data is a critical part to that because without proper data and enablement of that decisioning it's very difficult for any of us within the organization to achieve this goal or our customers so with that in mind we looked at how we need to organize this going back to Thinking Beyond and thinking ahead we don't want to be siled or short-sighted with our needs now rather than how do we put something that can scale as we scale our organization grow or products suits as well as customer base so these are some of the examples on the left side you could see that these are the different types of decision or use cases our users internal uses are trying to make from a few examples which segment of the customer is going to give the highest lifetime value which segment we should Target which campaign is working well if coming to fraud what is the chance potential chance for something to be default some customer to be default what is a customer risk profile there are additional fraud cases where we want to understand defraud patterns we also want to minimize false positives so these are the set of decisions we could make with data but then if you look how are we going to make those decisions there are three ways at least um in my opinion human intelligence you could get access to data you could look at the data analyze it make your decisions through reports Etc the other way of making decision is actually you have that predefined algorithms you implement that and that algorithms could make decisions on behalf of human third is most interesting and you might have heard that last yesterday and today a lot of AI right AI gen AI but there are multiple variations of that so there is like Predictive Analytics which is widely used for decision making and now there is Gen AI which is kind of like adding a different flavor which is making more human machine collaboration so you you have these are different ways we want to make decisions so when we want to make these decisions how we Design This one how do we enable this that's the next question so this is how we are looking at genius if you think Beyond maybe when you start it is easy for one organization to do all the Data Insights for in their organization but it's not scalable so kind of mentioned Marketplace this is kind of a Marketplace person so you want to democratize data what that means you centralize your data in one place both external and internal but then you decentralize the consumption of the data but it needs to be governed well you need to have God rules you need to ensure data quality that kind of like outlines the importance of data quality you want to ensure proper access control for that data you also want to enable flexible analytics because when you democratize data based on the authorization or or the need they should be the user should be able to get the data into a local environment or their own work area means not not the machines like in in an infrastructure where they could actually analyze the data they could actually model the data and use data to make right decisions so now with all this in mind this is what we have done a genius they decided to build an ecosystem that supports data centralization and also that can support decentralized consumption we haven't built all of this yet we are in the path of building some the centralization is almost done decentralization is actually being built as we speak so if we look at some examples I talked about loans we have multiple sources where we get the data our core banking underwriting App application uh processing components we get data at customer level we get data at account level we also get data at transaction level we want all this data coming to a landing Zone one place then sequence of processes like security encryption cleaning aggregation transformation and then the storage we store it to two different formats like one is row file format at our data Lake and the other one is actually more tabular format uh using bigquery in our data warehouse that is more accessible and also more domain driven we are using pretty much Google ecosystem in this process because um when we when we looked at different options we found this ecosystem probably meet our needs and we can scale and we could consider other tools if we need uh if we see a need in the future but right now we kind of like pretty much went with uh full um full of Google set of tool set um one exception probably data governs governance we are using collibra because that's kind of our standard across the board but it's well integrated with the ecosystem but the idea is actually like once you have Central data through proper governance quality you enable this through to different type of uses like it could be a leased environment it could be their own for example credit team could have their own environment where they could actually like get the data analyze the data build their credit rules experiment and then push into production they could visualize that we have looker so looker will enable you to visualize the data we do have a Vertex AI platform that eventually can be used to enhance your modeling so we thought like this type of an environment is easily scalable and also can support today we have loans tomorrow we have savings and we are going to have credit card different different products when we add over the years this model should be able to scale and support our needs we probably will make uh adjustment as we learned but the idea is actually how do we put together a vision where we can March towards so One Challenge we experienced to us like it's a different problem it's a good problem to have many organizations are looking for transformation for us a different problem we need to build it from scratch but when you build this from scratch the challenge is you're building the organization as well then you don't you know you're gaining the skill set who knows the data and also who can build the platform right this is where like we used lot of partnership and Professional Services to accelerate those development that really helped so we blend our internal team with industry expertise and kind of like build one cohesive team that can actually help us with the build out of some of these building blocks which kind of like pave the road for the future with that I will conclude my session and I think it's it's been an amazing journey a genius just want to make a point that we are hiring we are looking for a lot of expertise in Google because we are committed to investing Google so I'll call pankaj for uh closing thoughts thank you yeah uh thanks Anish thanks so um I think uh there's generally a lot to absorb in a session like this and I I think let me just simplify and just bring those analogies back in again so there are two things if you if you need to remember from this session two things think ahead think Beyond think ahead project talked about think about observability think about Fin Ops think how you're going to do user adoption very very important and then Anish brought it to life on how you could actually bring this all together to create the business impact you need to create so you need to actually think about you know what's your use case how are you going to use the data platform you're going to set up and how will you actually uh turn that into meaningful action so those are the two things you need to remember so if you're actually climbing a Mount Everest celebrate when you get to the Mount Everest but also think how you're going to get down what the next steps are so that's think ahead think Beyond you know we do a lot of learning from biomimicry I think think about a dragonfly think about every aspect of your data to decision Journey not just the central data warehouse or the data platform you're setting up [Music]

2023-12-26 04:54

Other news