Evan: Hello and welcome to the Minding Your Own Business podcast. I’m your host everyone is today. I am super excited to introduce Korri Jones. Korri is a senior lead machine learning engineer and innovation coach at Chick-Fil-A, home of the chicken sandwich. And what a cool-sounding job title that is. Prior to Chick-Fil-A, Korri worked as a business analyst in health care. He’s been a teacher. He’s been an adjunct professor. He’s super involved in the machine learning space and reality in the community writ large. He was part of the inaugural 40 under 40 at the University of Tennessee, where he has an MBA. He’s won a Volunteer of the Year award with the Urban League of Greater Atlanta. He’s been the recipient of the Looking to the Future Award and a ton of other things. But I want to be able to ask you some questions today so I won’t read through all of his accolades. Korri, welcome to the show. Thanks so much for joining us today.
Korri: Hey, it’s a pleasure to be here with you, Evan. So really stoked and excited.
Evan: Okay, awesome. I gave a high-level brief background. Korri, can you give us a little deeper background about yourself and sort of how you got to Chick-Fil-A and into the role that you’re in now?
Korri: Sure. And so currently I am, as you said, a silly machine learning engineer. I mean, some of my premier marching orders are really to think about this concept of model operations or machine learning operations, email ops, and starting to put some of the pieces in place to help unlock that overall potential for the business. And so that’s kind of like my current seat on the bus, but it’s kind of funny. We talk about how I got to Chick-Fil-A when I say it was just one of those random occurrences in my health care role. I was essentially a business analyst and I had worked myself out of the job. I think that seems to be my theme in life. And I remember sitting there and the company leadership said, Hey, we want you to be promoted to a principal consultant and you’re going to move from a data analyst to, a personal consultant, working directly with our CEO. I was like, yo, like this doesn’t happen. Begin to start moving places looking for backfill all of the things. And during that time I bumped into Chick-fil-A and I wasn’t looking. And this is kind of funny. It talks about those small touchpoints in life. They had a little book on the Chick-fil-a table at a conference, the Inference Operations Research Conference. And I walked over to the table because the book rhymed. It said True it. How’d you do it? That is the only reason I walked up to the table. And then from there, I met some amazing individuals. The interview kind of went well and the rest is history. And I ended up joining Chick-Fil-A and I’ve been here for about five and a half years.
Evan: That is fantastic for all the folks out there who are going to be a vendor at a conference or sponsor a conference that those titles are important. That signage is important. Maybe, a rhyme is a key there. Yeah, that’s it’s incredible how fortuitous it can be. And so I think it’s certainly been good for Chick-Fil-A. Hope it’s been a good, good decision for you. I want to ask a little bit about your role specifically. I think machine learning engineer, it’s probably not new necessarily, but it’s newer than I think more common terms and common roles, like a data scientist or a data engineer. Can you talk a little bit about where you sit in the organization and maybe how you interact with those roles that maybe are more familiar to folks? Like how do you interact with the data scientist? How do you interact with an IT department or a data engineer?
Korri: Yeah, no, I think I think that’s great. Yeah, I’m still an engineer. It is somewhat of a newer title, but it’s something that people have done. It has a lot of connotations depending on the organization. In some organizations, the Machine Learning Engineer means a full stack data scientist that others specifically within Chick-fil-A. I’m kind of like a hybrid. I have this understanding of the statistical underpinnings of the work that our data scientists are doing. I am not a data scientist. I do not build models. I just kind of clarified that. But I also have a deep understanding of data engineering. Talk about data governance, you talk about all of those things and so you smush both of those together. You get someone to like an email engineer. My seat in and of itself is a little bit unique because I leverage a lot of internal relationships as well as I bring a technical savvy kind of perspective to some of the things that I do. And so I interact with our data scientist quite a bit. I’m actually on the same team as our data science team, and I am like their champion. I’m the champion for all things engineering. If there’s some piece of infrastructure that is like given a lot of pain for a data scientist to be able to do their work well, I work with our partners in it and we help to come to remove that pain because at the end of the day, my, my, my big thing and this is just Korri speaking now is I want people to do what they are hired to do and really just crushing. I don’t need a data scientist doing all of the engineering work. I don’t need the engineer doing all the data science work. And so what I do is I make sure data scientists know what is in the way of you doing what you were hired to do? How do we make it so that you can flourish and show off those great skills that you’ve had to build up over the past couple of years? And that’s really how I interact. A lot of negotiations back and forth. I get into the weeds. So there are a couple of things I got no problem with. Open up an idea. You get to coding, put some building pipelines, whatever the case may be. And so it’s kind of a weird hybrid. I sometimes joke and say I’m like a middle child, you know, I’m not the oldest or the youngest. I’m somewhere in the middle. And so I can kind of flex into either or base off of the needs of the business.
Evan: Okay. That’s awesome. Now we talk with data scientists a lot, and that’s almost universally the biggest headache of trying to get the data, reclaim the data, and access the data. So having that bridge and advocate lets them do the things that they want to do and that they are most skilled at doing. I also mean, like you almost mentioned, is a bridge sort of between that i.t. Data engineering and data science. The book to this podcast is sort of named informed after Mining Your Own Business and talks about bridging the gap between data science and the business and big business decision-makers who would sort of rely on the outputs from some data science model. Do you have much of that in your role or are you just between data scientists and data engineers, or do you have links within the business with people that maybe aren’t technical at all but will rely on some of the data science output?
Korri: Yeah, I think that’s an amazing question and I do have links to the business and so that sometimes I’ll be a and so I think it’s also something unique about Chick-fil-A. And one of the reasons I’m always like, man, this is a sweet deal for our data scientist. It’s like they’re not necessarily just stuck in a corner coding, doing models all day. It’s like a part of the entire business lifecycle idea has come up. There’s a need for data science and they become, come and see me on those product and project teams. And in some of those cases I’ll also be aligned with it. We’re talking about data, unlocking data. We’re talking about cleanliness. What if there is something that’s somewhat sensitive? We don’t do a lot with that or just because that’s just too sticky. Why set yourself up for failure unnecessarily? But that becomes something that I do get an opportunity to do and I thoroughly enjoy it and I should do a little more of it with my innovation coach side of things, which sometimes I’ll get up and I’ll say, Man, I don’t know what today is going to look like. And it just brings joy to me because I get to bring that innovation mindset. So I also help out with some organizations as internally as we go through this process, to think through from IDEA, you have an idea, there’s a data science potential component to this. Let’s just talk through it. I’m not one of our analyst consultants. We have a team that does some of that. They help walk through the business process. But I’m like, I get a chance to walk alongside them and help out. You bring that technical expertise to the table.
Evan: Yeah, that is awesome. And I did want to get into the second latter part of your job title there as innovation coach. If we think a machine learning engineer is like a new or relatively unique job title innovation coach, really it’s impossible. It seemed impossible that machine learning engineer would be the least interesting of your two two job titles. But yeah, I mean, you just spoke to it a little bit, but could you tell us sort of in a nutshell what an innovation coach does and sort of how that role came to be? And, you know, just from a few minutes of chatting with you seems like you’re very well suited to fill that spot.
Korri: Yeah, another good one right there. So the innovation coaching side of my role was something I remember for a couple of years and about a year after I joined Chick-Fil-A, I saw that we had this innovation coaching community and I just saw all these great materials, your speakers, folks, and their mind was open. You know, they were thinking about how we prepare Chick-fil-A for the next 5 to 7 years? What do we think about this? You know, and also having that mindset of how we might think about a lot of the innovations that have come through our organization? And I was like, man, I want to be a part of that. So I sat down with people who were here at the time and I said, I want to do this. And I was interviewed by one of the folks who started the innovation coaching program at Chick-Fil-A. And it was really interesting. I was like, okay, so I’m getting interviewed for a side hustle within a kind of mini process. This for a second, but it was really good. And what I do there is I’m a resource for essentially anybody in our line of business. And what that means is if you have an idea, I can help you take that idea, flesh things out, ask them the right questions, even coordinate, facilitate breakout sessions in person, all of the cool things that you can do to help allow your mind to expand. And some people talk about thinking outside the box. We help to make sure that there is no box in the first place. And so you don’t have to try to think inside or outside of a box. Let’s step out of that and let’s create something brand new. And then sometimes we don’t create something brand new. We refine something that’s already there. Then there are many unknowns. And so I get a chance to do that all the way to launch from idea to launch. And so that’s pretty cool stuff. I’ve had a chance, to touch some amazing bodies of work in that seat of, my role as well as continue to kind of propel the importance of data science and analytics at Chick-Fil-A by engaging with some of our potential stakeholders who have been thinking about data science and then some of those that are current.
Evan: So yeah, that seems like a forward-thinking role and a really exciting picture here. It certainly doesn’t map directly to this, but we hear back in organizations there’s a digital transformation and trying to make things digital. But this feels a little more in the box, whereas innovation coach feels a lot wider decision space where you can sort of approach things. I’m curious if. If a lot of the ideas come from people that sit on your team in the data science team, or if they’re from other people around, just their ideas are somewhat related to machine learning or analytics.
Korri: Was kind of funny. So there are some cases where it’s on our team, but most of the cases come from the lines of business. So somebody I have something wild and crazy is like, well, I don’t know, but I’m curious. I want to see if this makes any sense. And so we have this concept of yes. And, you know, always a joke of folks is like there is no bad idea, there is none. Because what happens is that under the hood, people may say, oh, this has nothing to do with anything. We’ve already tried to stop in my seat on the bus to help them to unwrap and unravel them. So yes and yes, you want to have a giant blimp with Chick-Fil-A on it that knows to triangulate where customers are and it will drop airdrop towels with waffle fries in that area. Like you get kind of wild and crazy, but then it’s like, yes, and then.
Evan: You’re cross on that.
Korri: One. We can also have little airplane cows that go around to schools and drop off stuff and be like, okay. Yes. And you keep kind of growing these ideas. And so some of them deal with data science. The majority of them don’t. So I think that’s the beauty of the seat on the bus. It allows me also to expand my thinking by stepping outside of my realm of expertise like technical, and tactical expertise. Sure. And lead me on the innovation coach side. So that’s really where that’s how a lot of different functions get a chance to leverage us as innovation coaches.
Evan: Okay. Wow. Yeah, I like that idea a lot. And so hopefully, folks can I don’t know, it takes an innovation coach role to start the innovation coaching program. But hopefully, people can try to leverage that idea and sort of get some creative thought going. Okay. I want to circle back to two boring old machine learning stuff again. You said you don’t build model data. Scientists are going to be the ones to build a model. You help bridge that gap with engineering. And if you know a lot of what data scientists do, you know, may not be production work or production-focused work. Maybe it’s just like a quick one-off, but maybe a proof of concept, maybe just some exploratory analysis. I’m curious when you would get involved in sort of the work that a data scientist does or does it take until, hey, we’ve proven this out, this is we can implement this at a small place and now we need to take it to scale. When is a machine learning engineer? Need to come and help out with that.
Korri: I think it depends. That’s the MBA answer, right? It depends. And so there are cases where maybe pure experimentation, but there might be some questions with regards to how we best bring in this type of data necessary. The enterprise data we have like a team that helps out with that on enterprise analytics is like, okay, this is the right data, these are the things, but maybe there might be something, hey, the type of data I’m bringing in and say is the major, let’s say it’s unstructured data I’m trying to have the right compute is going to be able to support this. And so I kind of step in and kind of help to think through that and put some framework on it. Because sometimes, again, let’s just be honest, when you’re working on stuff in the cloud, no matter how experienced you are, sometimes it’s like, I think this is the right computer. Let me try. And it, it’s just a whole bunch of optimization that has to go after that. But again, that’s normally a nitty-gritty engineering function and so try to try to step in and help out as much as possible. We have it as an organization. We have some amazing data scientists that get in there and they do some just tremendous work. And so I step in really to be any type of support. Again, whether it is bridging the gap, somebody, engineering teams unlocking and pushing data, or even just being the squeaky wheel in a room. I have no problem with popping up in meetings and saying, hey, you know what? This is a valuable dataset. And unfortunately, the computing medium that you have is not allowing us to run the models that we need. Is there a way for us to kind of enhance that, compute less leverage spark? I wish there were some other type of distributed computer outside it like, so I get a chance to do that. So it depends on some cases of pure experimentation data scientists rock and roll with that depending on the size scope of the data in a type of project and then your net positive permutation phase. But then there may be cases where, hey, they got everything, the field stuff is pretty good. They’ve done all, all the experimentation. They feel like they have an amazing model. And then myself or our teams, our partners in our IT group, we call it digital transformation technology. So you ever hear me say that’s what it’s kind of like, equivalent to it in other organizations? But, you know, we partner with them and we just make it, move it, get it, everything kind of wired up for production. And so production has a lot of connotations, but that’s a whole nother topic in itself.
Evan: Sure. Yeah. And I don’t necessarily want to get into the nitty-gritty of technical details on production, analyzing something that’s sort of played out and shown some value. But is there a way in this, you know, maybe. Maybe at scale, this doesn’t even matter anyway. Are you part of the decision to sort of prioritizing that? You’ve got quite a team of data scientists. If you’ve got, hey, this shows promise, this shows promise, this shows promise. There’s only so much resource to try to production-wise or scale-out some of those solutions. Are you involved in that decision all or how does that come to be?
Korri: I think that’s an interesting question because one of the things we try to do, you know, you hear that credo, principle, 80% a model, never make it in production. And so what happens is most of the work that we’re doing is based on business needs. And so it’s kind of like you’re working on something that’s going to deliver value to the budget because we have clear stakeholders. We have all of that. So odds are that the majority of the work that our team is doing will be put in some form of production and less during experimentation. Identify if you’re missing key data or if there’s something that’s not necessarily viable for us even to push any further with this body of work until blockers are removed. And so I think that puts us in a really good place. So it’s not necessarily data science, just kind of playing with data. See this data, I have this crazy idea. And so that sets us up, I believe, for success with that. And so with regards to priorities, prioritizing what goes into production when, let’s say, when you have like, let’s say I got 50 different projects that need to go into production again, that also depends. There may be certain teams of our data scientists to have like support from engineers within our DTT group. And so what happens is, let’s say those five at a 15, those first five have like your support group, those support engineers will help them push production. Then there may be cases where some of our data scientists are like, Hey, I’m pretty nifty with this stuff. I think I can get it to about 95% down from a production standpoint. I just need somebody to run through the code review, make sure everything is wired up appropriately, and then do that last 5%. And so there are cases like that.
Evan: Okay. Yeah, that makes perfect sense. And also ties back into your discussion before about being able to link with the lines of business and being involved in those conversations to sort of help push that prioritization through. Okay. So I want to ask you one more question here. It’s not always obvious. You know, maybe there’s occasionally tension between the data science team and the data engineering team or the line of business and trying to get aligned on which projects to pursue even as a proof of concept. I mean, which ones to scale out. Let’s just say we’ve got an easy button, we’ve got a magic button, and everybody is aligned to the Korri Jones vision. You get to pick something and everybody’s 100% on board. This is what we all want to work on, and we all share the same vision that Korri Jones has. Is there a magical fairy tale land? But if that’s the case, what do you want to push for? What would you like to do now?
Korri: And why are you hitting with these really difficult questions that are going to make me think hard so that if I had like that magic finance guy, again, all stakeholders are happy. Everybody’s going to be fully aligned. I have what I call a, b, b b picture audacious goal. When I think about model B impression production and being able to deliver continuous, consistent, and resilient value to the business, my idea would be I want us to be able to put 80 models after the data scientist and our immediate team has said that we feel like this is good enough to be put into production within 5 minutes. I want it to be a frictionless process with as much automation and processes underlying so it’s not necessary to bloat. So it’s like, okay, I’ve done my experimentation, this is where this goes, this is the next step. These are the next steps. It’s so crystal clear, so frictionless. Within 5 minutes, that data scientist was given a checkmark. I feel good. That’s all right. If my name’s on this, we’re out there, so I think that’ll be one thing. The other one I’m going to kind of channel for all of our machinima engineers and model operations professionals, honestly, just Chick-Fil-A, but all over the world. And one of the themes that I continue to hear is like the challenges where it’s kind of like engineering teams over here and then your data science team is over here. And as you said, it’s kind of like a back and forth and kind of working with them and then different people trying to work with the stakeholders. It’s just it’s like this big mess. If I had a huge vision, I would say, allow your teams to have like small teams, small but effective teams, where there’s a data scientist, an engineer, a data analyst for visualization, data managing whatever the case may be. I kind of have like small 3 to 4 man or woman teams that focus on these projects and then like they can do technical execution to have the right people there. And I think that would change a whole bunch. And or. I do, yes, and yes. And to begin to think about how you structure your teams. And this isn’t for Chick-Fil-A. This is for everybody based on a lot of themes that I’ve heard. But that be the other thing like I’ve seen some amazing work for those teams that have the engineers and the data scientist sitting on the same team, the same leadership, the same vision. And so it’s not like, hey, we’re over here in engineering. We don’t. And all this stuff or, hey, we’re data scientists. We all understand all this. Let’s get alignment. All your departmental meetings are at this time it conflicts in the meeting. Get all this back and forth. Yeah. One vision, one leader executed. And I think that would be another pie in the sky for me, we kind of have that solid here at Chick-Fil-A, but it’s just something I think about for a lot of the other professionals in my space.
Evan: Yeah, I like that vision a lot. And there is often a lot of friction between those engineers and the data scientist, even if it’s not like tension and negative friction. There are just things and communication and thoughts and stakes to try to overcome to push forward with anything. So, yeah, I think his teams are growing their analytic capabilities like most teams are. It’s really important to think about organizational structure, and that is a key piece there. How do you reduce that friction between those engineers and the scientist? Okay. I know I told you that was the last question, but I got to ask one more question. I’m a happy Chick-Fil-A customer. I love going to the store with a spicy chicken sandwich myself. I’ve had nothing but positive interactions and I don’t know, a thousand or a million visits to Chick fil A. Everybody’s so pleasant. Is it like that at corporate Chick-Fil-A, too? Is it like that on the machine learning team or are people just generally super nice and pleasant to be around a Chick-Fil-A? You know.
Korri: I think that’s an amazing question just because I always think about that every day. It’s kind of like you have to pinch yourself from a Chick-Fil-A standpoint because I’m like, man, these are some amazing human beings, folks who come in with the mindset of we’re going to have the best expectations, you know, or assume the best. So I may be having a bad day and I might say something maybe not as kind. But we also always assume the best. And we have conversations that afflicted many other organizations. And just like I’ve had discussions with leadership and just my peers that I would have never dreamed about. And so ultimately. Yes, yes. I’ll tell you a funny story. When I was interviewing at Chick-Fil-A, I remember we had a bunch of different companies as well. I mean, it was like it was that season. And so I would go to these organizations and you see people looking down sad, looks like they just wore out at these different companies. But I walked into Chick-fil-A. I didn’t feel any of that negativity. I didn’t feel like folks who were like, I signed up for 40 hours. I’m working 90 days all the way to get my job done. I didn’t see any of that. Like everybody was kind and nice. I’m sitting at the front. You know, in retrospect, it was like senior leadership was stopping by and they saw me sitting up there in my seat, you know, it’s like, okay, sure, I’m good. I want to put my best foot forward. They’re like, sort of you’re like some coffee, some hot chocolate like this. This wasn’t like our concierge desk. They checked on me, but these were just people walking in. And that heartbeat of servant leadership from day one was one of the reasons that I was like, I think I think this is a place I can be at. And those were some of the things that made a difference to me. But yeah, it was. The people internally are amazing and I’m just so grateful to get the opportunity to work with and for a company that respects me as an individual, but also that makes sure that we have the right heartbeat. And from the top-down, I mean, I got all types of stories from a previous CEO, Dan Cathy, and a current CEO. Andrew likes just, just like interactions that we’ve had, whether it be cracking ball jokes or are just, you know, just talking about life and business. And so it’s amazing.
Evan: That. That’s great. That is it. That is a great answer. It makes Chick-Fil-A sound great. I’m sure Chick-Fil-A is super happy to have you there. Korri Jones, We were super happy to have you here on the Mining Your Business podcast today. That’s all the time we have. Korri, thanks so much for joining us.
Korri: Pleasure is completely mine.
Evan: All right. Fantastic. If you’re looking if you’re a data scientist, data engineer, or machine learning ops person. Chick-Fil-A sounds like the place to be. So stay plugged in there for sure. If you’re looking for podcast material, make sure to like and subscribe and multiply and divide. And we’ll see you next time on the Minding Your Business podcast. Korri Thank you.