The Data Hustle — Thais Cooke

>> Yeah, sure. Thank you for having me. And um yeah, so I ended up in data almost by accident. So I'm going to try to make

this very quickly. I'm a dentist in my country in Brazil. Came here to the United States, became a dental hygienist and I was practicing for about 10 years in the clinical space and I was already very burned out. I say that clinical healthcare is not for the weak. It's a really really hard job to do. It's burned out for many different reasons. Not going to go into details. And I knew I needed to do something else besides that. And then co hit and I said, "That's it. I'm tapped out. I don't want to do this anymore." And took a step back, took a breath. And a friend of mine at that time, she contact me. I, you know, was talking to her. We're

going there. This was 2020, end of 2020. So, start doing that. And after a while, I said because I'm a geek, I start noticing some patterns in the data. And I start seeing like, oh, you know, this month we have a lot of implants. Oh, this month we have a new office is showing so much this month. Oh, last month we didn't see this office so much. I wonder what happened. So, I literally went on Google and type how to analyze data. >> I had no idea what I was doing. Had zero clue. And I start to look into YouTube videos and everything. And one of the videos um one thing that was very clear to me was you have to learn SQL. That

was it. You have to learn SQL. You have to learn SQL is the bread and butter. And I decided, okay, I'm going to try learn this SQL thing and if it works, great. If it makes sense, I'm going to pursue that. If it goes completely over my head and I see it's not for me, then it's not for me. And I say that when I wrote select star, from that moment, I fell in love with SEO right away. It just it just made so much sense and I was like, "Oh my god, I can do all of this with the data." So, make a long story short, um got better in SQL, was waking up early every day to study study

started know coming through all of these and so I kind of sort of became a data analyst on the spot >> right there. uh but this was it was a small company they didn't have a data department like I said they didn't have have a data warehouse nothing like that and so eventually when I start feeling my my skills are getting better I'm like okay I need to you know do something about it I think I'm ready for you know a bigger job and start applying and since I had this healthcare background I decided to use that in my advantage start to look for healthcare data and these jobs and ended up where I am Now that we deal with behavioral health and

you know overall health as well. So that's it. That's my story. >> Interesting transition. You became an analyst after working with data. >> Yeah. Exactly. Analyst by accident. >> It's funny how some people come from the technical domain and then they need to learn like whatever business they're working at and some people that do the opposite and then somehow both of those people tend to converge in the role. It's nice to see. Yes. Yes. It's true. And I said that's the beauty of the data industry is that we have people from so many different backgrounds. So we look at those problems that we receive from so many different angles and it's it's really it's really rich to to do that.

>> Yeah, I think so. the old way that the data analyst used to do the work which is around the time when I did this transition into you know the data space I mean I actually just the irony of the matter is when I got this you know um this role that I am now it was November 2022 same month that chat dpt came

mainstream so you can imagine and then you started the news like we don't need data analysts anymore I was like What? So hard for this. >> So you can imagine my reaction, right? I mean >> the irony. Oh my god. But but yeah, but the way that I seen up to them, what I was doing is I was collecting requirements from the stakeholders, build SQL queries from scratch based on the logic. You spend a lot of time cleaning and formatting data. And I remember like even when I was doing this transition, I had some friends who were data analysts and they say, "Oh, majority of your time you're going to spend cleaning data. That's what you're going to do most of the time." Which was

true, you know. >> And so now, of course, we still have to clean data. It's not like we don't have to, but that is done so much faster. So I think what is is slowing is slowly dying let's say is those like to be so focused on on that and nowadays AI can build your SQL based on the logic that you give of course don't get me wrong I'm not saying that you don't have to know how to code anymore it is important of course you need to know what your query is doing you need to know if the query the AI is giving you is the best way to get to the problem that quer is optimized and all of that But what is

being cut right now is those writing from a scratch type of of work that we spend so much time on that. So not saying that this is not important but it's just the time that we spend has shifted more into collecting the requirements working on the logic working with stakeholders. So that's what is being reborn. So it's not that data analysts are going away, you know, we're being replaced. It's just the time we spend on our tasks is shifting. So we're shifting more into context and all of those things that AI still can to do for us. kind of crazy how AI like brought some balance but a bit unequal because it does help you a little bit with the

domain expertise like you can still ask questions and get it >> but it feels like when you're an analyst the amount of things that you should know in technical terms or that you should be able to execute has increased a lot because they made it easier >> but when it comes to domain expertise is like not growing in terms of how much you're expected to be able to do equally. >> Yeah. >> Have you noticed that? >> Yeah. No, I agree with you to know it's not about only using AI is knowing which problems are important solving >> and and if AI is even the right solution at the moment. >> We don't spend like 80% of our time cleaning data, but we spend the cleaning

AI output. >> That's true. >> Yes. >> It it does hit to an interesting question though. You know, you spent all of and your initial response to getting the job and chief coming out kind of fits on that vericular. It is we spend all of these years, Bill, on these implementation skills and then you have like five or 10 years of these skills and suddenly it feels like that's sort of taken away from you but at the same time as one was indicating we're expected to do a lot more and the same role with the same amount of people. Um, how have you experienced that like on an emotional side like going through this uh AI roller coaster? >> Well, it's it's so weird because it

seems that people are having this fear of missing out with AI >> and there is I mean lately not as much. I think now the conversation start to get a little bit more grounded, thank goodness, into what AI can and cannot do and how to use AI in a in a meaningful manner and in a way that it's not just oh spray AI and you know and that's it just give to AI to do it and and you know and verify and you're done. It's not like this the system the structure the domain knowledge of the context of this need to be in place before we put AI in there. So and and like I said sometimes AI might not be the solution

for the problem. Sometimes it might be a smaller problem. We can do something you know simple manual be faster than than using AI depending on the problem of course. >> I've personally I don't know about you I've caught myself asking and I learn something that was objectively quicker to Google. >> One thing that we really need to be mindful of. >> Yep. So you do kind of hit on some of these topics like we have to understand how to deal with context. U I kind of want to dig a little bit deeper into this. So >> obviously we all kind of have seen now uh AI in the wild for a couple of years and um where the industry uh seems to

lost these words. critical thinking >> to know so obvious but I lost the words but the critical thinking to to know if what AI is giving me is actually what I'm looking for and that just comes if you have the domain knowledge to do that if you otherwise you're going to believe whatever AI says and you're going to copy and paste there but yeah but when it comes to judgment um for instance this um this actually uh happened to me. There was a this was a while ago um a stakeholder came to me and wanted a dashboard and wanted everything under the sun in the dashboard everything you can imagine like all the questions and I mean it's easy for us to think hey I

What what are the questions? What is the priority? What is really important that if you open this dashboard is the first number you need to see and then that whole list start to get very small and then it start to be

happen to be an analyst in an industry where the consequences are very extreme like we're talking for finance that sometimes means like fines that are astronomical in healthcare. I mean you can hurt somebody if you do an incorrect joint. So being already in a field like that trains you to use your judgment a lot more and I can imagine that for somebody like you taste that has been working in a field where like you there are numbers but indeed there are actually people behind it >> you to think twice. So it's interesting that also that difference between fields. >> Oh yeah absolutely. I see a difference like and sometimes I read blog posts from people that work in tech that is so

forward and I'm like why I'm not doing this? I feel like I'm missing on it but but I understand that healthcare we do have you know because the consequences are so much so much more that that move fast and break things works in certain fields but don't break stuff in healthcare that's not that's not going to be that's not going to be good >> oh no I mean that that that is definitely a nogo in healthcare we do not we do not want to hit on that uh >> work style Um, >> nope. >> So, from a governance perspective, obviously you have a ton of experience dealing with u governance in healthcare. You even have a LinkedIn learning course on this. Um,

Because something I've noticed talking to uh folks in different healthcare companies implementing AI is that there is sort of this restraint to use it to its full capacity obviously permanently so because of the risks associated to it >> but how do you balance you know let's say a um CEO or CTO saying I want I want to reach for the moon uh but do it safely what is your perspective on that there? >> Well, there there's the parts of uh governance in healthcare. There are non-negotiables and I mean, of course, there's the laws and regulations that we have to abide to like in the United States, we have HIPPA

laws, we have, you know, all those laws that we have >> for obvious reasons we have to abide to. >> Those are the non-negotiables. So, those are like, okay, put a lock in here. Non-negotiable. we have to abide to this because that's the nature of the industry you know we have to do this and there is also the but then if you flip the coin um there's the consequences of I'm going to build something that has high consequences for my patient sure put all the safeguards in there but at the same time you don't want to fall into that oh there's a query that is going to show something that really doesn't have much consequence Let's say I don't know track appointment times

just to you know something that pops to my mind. >> That's you know a non-emergency appointment time is not life or death. Yeah it sucks to be sitting there waiting if that's not correct of course but it's not something having a heart attack. Let's put this way. So I don't have in this context of like something that there's a consequence there's not that much to oh we have to check this query and do three checks and have compliance check on this before it gets approved before no that one you can move fast you can move fast in those things but when there's something that is either the law you have to abide otherwise it's going to be a nightmare or something that has really high

consequences those two things focus put your governance in there. The other things that don't need to escalate, then move fast there. You know, move fast, make your mistakes, go back, fix it. But so it's a balance in what really need to be restrict and what you can you can just, you know, not lose a little bit. >> That's a great point. I don't think enough data people are dividing their data products, for example, by the consequences of something being wrong. for the amount of governance it really needs. I think most organizations try to keep it uniform >> maybe because they don't want to have that discussion of what is more important what is necessary but >> doing it is very powerful and I guess

Uh, and what would you advise to them to do differently?

>> Well, I would say, believe it or not, just have empathy and try to understand the world a little bit of what they're trying to accomplish. um what is important for them and speak to them in the language that they understand instead of doing this like technical jargon. I think sometimes we we fall a lot into the into using technical jargon and and things like that. So for instance, if I say to my stakeholder, oh the data quality is just not there. Okay, sure. They're not going to understand what it is. What they need to hear is hey guess what we don't need we don't have a share meaning on this metric so we need to define this before we build anything so we are all on the

same page because this is what you want and this is how we're going to accomplish okay sure they will they will go for it but if you just say oh no yeah we don't have data quality you need more investment that all right they're going to sit there like okay >> raises an interesting point which is trusting data is not an inherent property of the data but it's also the part of the relationship of the people that build the >> data assets and the stakeholders. >> Yep. I mean talking about translating things obviously we have had a ton of experience in the field of translating business inquiries to um answers in our data models or data warehouses >> and so forth.

Do you think it's it's effective? Do you think we're there or do you think this is going to require a couple more years of effort? Um, because from my experience in in healthcare, there's certain aspects as you mentioned less critical, more straightforward operations. Sure, I can see that. But there are definitely some very complicated backend things like payments and claims and um a bunch of other highly highly complicated healthcare systems where this might be much more

>> Maybe to simplify, I'm curious like which are the things you're happy you don't have to do anymore because AI nails it. What are the things that you're probably gonna wait a few more years before trusting AI? >> So, well, I'm an optimist. So, I think when it comes to semantic layers, I it's very promising. I I you know, I see and I see in in tech and I'm like, "Oh my god, this is amazing. Amazing what I see." And I keep thinking about how to apply that to healthcare. But as you said, yes, because healthcare data is so fragmented and you know, and then there's all the

different definitions and the meanings and the share definition. So I don't think we're quite there yet in healthcare, but I can see a lot of promising and again goes back to the consequences. So you're going to put an agentic workflow running on something that has not been defined well defined in healthcare and those definitions in healthcare are so hard to bring it together. >> So this aent uh this identical workflow is not going to solve the ambiguity. It's just going to escalate. So I can see like healthc care analytics shifting a lot towards the the the you know which is funny because it goes back to the same problems we have always had a line on definitions and everything and then

you put AI on top >> but yeah healthcare is going to be a challenge. >> Yeah just AI is handling some of the technical complexity so the other problems become more visible. >> Exactly. Cool. Which that's kind of a very nice bridge to all of the questions that we have and that is we're curious about your thoughts and your ideas of how the field is going to change in the coming years. >> Oh, how it's going to change. You know, it's so hard to make a prediction because it's changing so fast. But I think um the analysts are going to start um at the goal of like what is changing right now is compared to what I said before. We used to start on the query

and just you know write and give this but I think now we're going to really have to focus on the goal for analysis the decision we're trying to support and work backwards. We starting there spend a lot of time there focus on there. Okay, we nailed the decision. We know how to keep track as reality changes. Good. Let's now move to the definitions. How we're going to define that? So I think that's how is going to be the shift and then we touch the code. So I think this pre-work that comes before we start coding. I can see the industry moving a little bit more towards that. But yeah, so I think the next one to five years maybe that's what that's what

it is. I think everybody wants to do AI now for I know for all the reasons. So I think now we're going to be on okay great everybody wants AI but how can we implement this in a way that is actually going to bring return to us that is actually going to be meaningful. So >> it's interesting because some some of the approaches like uh first focus on value or first ask the questions or those that advice has existed for a long time but I think now with AI I notice it's becoming a same with test before write unit test before you start coding a lot of that has become a bit easier now that we have LLM. So maybe a natural

also when you're doing this, try to see try to play with AI when you are, you know, I'm trying to picture myself when I was doing my career transition that was preai. And if I had AI in my hands, I would probably be like, "Okay, this I can do by hand. This is good. I know SQL. I know, you know, how to write this." Oh, maybe I can help him in this. And play around. Play around with AI. You know, get your hands dirty. Data is you have to get your hands dirty with it, AI or not, and see how that's going to bring value to you. But definitely get comfortable using AI. uh even if it's for projects you know um that you're

going to do in your personal time and but yes use that definitely >> I agree it sounds like a solid advice >> to be honest all sorts of things are going to change even what's our definition the fundamentals might change >> but I think right now getting comfortable with AI it's going to become a really good comp it's kind of like the 20 years ago learn Excel and if you are applying for a job mentioning your CV Yeah. >> Know how to play with AI is the new learn Excel. >> Yeah. I'm I'm old enough to remember when my school first put a computer lab. We didn't have computers. >> Same here. >> And there was all those rules. I was in

middle school, I think. Yeah. I think it was middle school or Yeah. Or or junior in high school. And and there were all those rules. You know, the computers are like huge and what you could and you couldn't do around the computer. Don't bring food around the computer. Don't I'm old enough to remember that. >> But we had to learn. You had to learn. I guess something that I just thought of is I I guess a lot of people coming into the field or even people that are meteors or seniors trying to break out and do their own thing are sort of feeling the pressure and and and feel like isolating themselves away from from the issues. I I've noticed there's some

How should they think about building their brand? uh push putting themselves out there and sort of building this credibility uh and relationship uh with the industry, with individuals, with prospects, so >> they can really set themselves up for success. >> Well, so it's funny because when I first start on LinkedIn, I literally just went there to talk to people with zero intention of like building a brand or

know reach out to people and say hey love your piece that you work on this I have some ideas you want to talk and people very receptive so I would definitely start to do this uh online um start online but if uh I also now that we're not in co anymore try to find meetups in your area like meetups, you know, happy hours, you know, things like this and just go there and talk to people, brainstorm together and you have a thought, post online, you know, and and comment on other people's post too. Just just to have that, but not with the intention of, oh my god, I'm gonna build my brand. I'm going to be, you know, because people can smell that from far.

doing these things as much, but maybe they should a little bit more? I'm an introvert, believe it or not. >> It's very hard to look like, but yes, I am an introvert. >> Um, well, just just go and don't think that you have to be like the life of the party and everybody needs to look at you or anything like this. >> Go, you know, like I said, go to a meetup. just say hi to one person, you know, have one-on-one conversation or if you are more comfortable sometimes just observing, just sit there observe in a larger group, but little by little, you know, little steps. You don't have to be the life of the party. You don't have to

be all eyes on you or anything. I actually hate that, but but yeah, but just go and you never know who you're going to meet, you know. And it's interesting because >> I had met a lot of people who are extremely introverted but when you are on oneonone with them >> awesome make nice conversation and everything. So >> no I like to think of community involvement as you can participate in lots of ways from writing codes writing articles talking at a meetup. >> We can make a massive list. I like to bucket all of these things into two lists, which are the ones that come naturally to me, but don't challenge me, don't challenge me too much, >> and the ones that are not that easy to

me personally, but I still find it valuable to put myself out there. >> How does it look for you, T? Which are the things that come very naturally? Which are the ones that like not my cup of tea? And which ones maybe don't come as natural, but you're like, I still think it's valuable to push a bit here. that come I think the things that come naturally is like I I and I don't know if this is because I work remote and I go a little stir crazy at home so I like maybe that's why so I just you know just to go out and like in those like you say little communities I love things that are not vendor oriented all due respect

to vendors okay please all due respect to them but I like to go to those little community like to see what you know my peers are doing, what is working their company, what is not working. So I think that just comes natural like I for me it's like going to a happy hour getting out of the house um but I'm talking to people about data. So yeah, >> thanks. And some of them that might not be that easy for you personally, but you still think it's nice to do them to get out of your comfort zone. of things that believe it or not. So I I have spoken day-to-day Texas and I love putting speak you know speaking together and

everything but it is out of my comfort zone to be speaking like with an audience live audience in front of me. It it's intimidating. It's out of my comfort zone for sure but I do it anyways. I think I compared to like a roller coaster. You are, oh my god, I can't do this. And then when you're there, you're like, what am I doing? And then afterwards, you're like, okay, this wasn't so bad. So, I think for me to be invited to speak, I still get a little, you know, but I don't know, maybe everybody gets a little bit like that. Well, I do think it's a bit different for everybody, but it's it's nice to hear from your perspective how there are

>> Okay. So I try to see like why are they being resistant? what is you know what is really hey maybe maybe they didn't have luck with data analyst before they're like oh you

there is some sort of resistance. >> Mhm. >> But uh too also know that sometimes and a lot of times that's not even personal. >> They might be like disappointed with somebody else. It's your dashboard can be amazing, but they still have seen so many examples of bad data that it's >> Yeah. And honestly, I think mo most of the time it's not really personal. Most of the time, maybe it is, I don't know, but most of the time somebody that is brand new and you just match, why would that be personal? And on that topic of empathy and the skills that you brought from your previous job into your new data role, I do get this question a lot from people

teacher you were, you know, whatever you were, you had front row seat to which problems are worth solving. So that's that's a skill that you know if somebody comes from day be like I've noticed this because you know I came from dentistry and I switched to behavioral health it is still a little bit different like I still had to there was a learning curve on what behavioral health is versus dentistry they are in healthcare both but it's little bit of a different angle. So, but I see people that work in behavioral health and be and came into data. Oh my god, they know this like like this like what is important, what is not important, what is priority, what is not, what the

problems worth solving, which you know problems are just noise. So when you you do this uh imagine almost like you you know let's say you're a teacher you have this basket of skills and now you're going to go into data. So what is important to bring here? So you're going to take away the skills that are not important and focus in the ones that this is going to help me solve your problems and you add the technical skills there and here you go. You can walk in data because you know the problems that you're going to solve. You're just going to use data to do it. >> So put yourself as a problem solver. >> Yeah. You need to of course accept that

You know, I just read an article recently about very exciting things that are being done in remote areas in the world. >> That those are really big issues that we don't like to use first world problems. We don't think about that, >> you know, and to see the technology is actually helping that. So I saw something I don't know exactly. I think it was in Africa like um to detect um spots of malaria something like this. I'm sorry like I don't remember exactly the details but to detect spo sport spots of malaria. So what to do to eradicate that and you

know using technology to track that some something like that. So what excites me is to see those problems that are bigger than us like that we don't think about but those problems and you see that wow we're using technology to solve that you know that excites me like when I read things like this I'm like oh my god this is amazing this is amazing we're doing something right there is a lot of hype there's a lot of oh it's going to take everybody's job and this and that but there's a lot of good that can happen too >> that is a wonderful way to look at it. >> Yeah. Yeah. >> I remember when I first read I was at

>> Okay. So I'm on LinkedIn. I'm very active there. I am moving a lot to

>> Excellent. The the the Texas tradition is uh is kept. I I do like that. U >> Yeah. >> Wonderful. Okay. Um >> yeah, >> I think you can um call it quits here.