The Happy Engineer Podcast

067: How Data Science and Machine Learning Lead to Optimizing Your Life with Kristen Kehrer

What does the future of data science and machine learning have to teach you about life?

How can you take the best parts of your work life and use them to optimize your home life?

In this episode, we follow the data and learn as we go with Developer Advocate at CometML, Kristen Kehrer. She has been awarded “LinkedIn Top Voice” in data science and continues to share remarkable content with her audience of over 88,000 technical leaders.

Her passion for machine learning is mirrored by passion for optimizing her life, and we explore both together today.

Kristen is a former Data Science instructor at UC Berkeley Ext, Faculty/SME at Emeritus Institute of Management and Founder of Data Moves Me, LLC. Kristen holds an MS in Applied Statistics from Worcester Polytechnic Institute and a BS in Mathematics.

So press play and let’s chat… it’s time to take a look at the data of our lives and iterate to the next level!


>> Then join The Happy Engineer Community online and get access to bonus content and coaching in our free group >>



The Happy Engineer Podcast




Listen on Apple Podcasts // Spotify // Android // iHeartRadio


Previous Episode 066: Q&A #3 with Zach White – The Best Morning Routines for Engineering Leaders and Advice for Student Engineers





There is one simple but important point that came up early in my conversation with Kristen that I want to recap with you here. 

In the beginning of our chat, Kristen made a comment that there was a point in her life where she became obsessed with the idea of optimizing her life. 

I love the parallel between how she approached optimizing data systems and all the technology she worked with, and her approach to optimizing continuously learning and taking new actions in her life.

This leads me to my challenge for you today.


Curiosity is one of those traits that engineering leaders are naturally drawn into, especially early in our careers. 

I want to challenge you to tap back into that part of you today and tomorrow. 

In every situation, curiosity can be a really powerful energy, attitude state, way of being to come back to.

When you’re furious, get curious.

I say that not because I think you’ve lost it. 

I say that because I know firsthand how the bureaucracy of the business that you deal with and all of the non-core engineering. 

I get it. I’ve been there. I’ve built my career in engineering. I understand there’s a lot of things that we must do in our jobs that are not aligned with our deepest passions, for the technology, for the products, for design.

I’m not saying that it’s a bad thing that we have non-engineering work as part of our roles, but what can happen is we forget to tap into the part of us that drew us into engineering in the first place. 

So I hope you had fun today and go get curious, be a little kid again, be like that kid in the sandbox, who’s just playing and having fun. 

The sandbox of your life is right in front of you, right? 

So go out and have fun. 

Be curious, learn something new today, not by scrolling Facebook and Instagram, but by digging into something that genuinely ignites your passion for life, enjoy doing it. 

And along the way, don’t forget to crush comfort, create courage.

And until next time let’s do this.

Question #2

From Maria, student of engineering.

What advice or tips do you have for those of us still in college? Is there anything that you would do differently based on what you know now?

1) Reconsider just going after the grades

Something I wish I had done differently is not focusing so much on always pursuing the high grades. 

While that served me well in some ways, it also left A LOT on the table because I could have used more focus on really understanding the principles and how to apply the material.

In summary, allow yourself to fall in love with the principles and the material. 

2) Go to office hours. 

I see so many engineers nowadays that are afraid of speaking to superiors. At school, office hours are your opportunity to practice just walking in the room with somebody who you see as a superior and realize that you just can have a chat.

Don’t be afraid to hang out with your professors. 

I wish I had done more of that because it helps you discover how all that you do as homework actually plays out out there in the real world. 

3) Fewer groups, bigger impact.

Back in my Whirlpool days, I did a lot of hiring of college grads. And one of the things that I saw a lot was resumes packed with activities.

And here’s the deal.

If all you’re doing is showing up to a bunch of parties and member meetings and different things for all these groups you paid your dues and you got your permission to put it on your resume, but you didn’t actually move the needle on anything of impact in those organizations. 

To me, that’s not impressive.

Instead, show me (imagine I was to hire you) something meaningful you accomplished. Something that came as a result of you having put your heart and soul into it.

4) Look at yourself as a whole person. 

You will not just be your engineering career. You will (or might) also be a parent, a spouse, and you will have interests outside of your profession.

At OACO we do coaching for engineering leaders around the concept of how to create career success without suffering burnout and becoming one dimensional, working too much, or really only achieving success in your career, but your marriage falls apart or your health falls apart.

So, if you’re still in your youth and these things are not an issue yet, my advice is to start looking at yourself as a whole person in all the different facets of your life, not just you the engineer.



Kristen Kehrer is currently a Developer Advocate at CometML sharing about the importance of reproducibility and model monitoring in machine learning.

Kristen was a LinkedIn Top Voice in data science and since 2010 has been delivering innovative and actionable statistical modeling and machine learning solutions across multiple industries, including eCommerce, healthcare, and utilities.

Previously Kristen was a Data Science instructor at UC Berkeley Ext, Faculty/SME at Emeritus Institute of Management and Founder of Data Moves Me, LLC. Kristen holds an MS in Applied Statistics from Worcester Polytechnic Institute and a BS in Mathematics.





Please note the full transcript is 90-95% accuracy. Reference the podcast audio to confirm exact quotations.

[00:00:00] Zach White: All right. All right. Welcome back Happy engineers and super pumped to be with you today, Kristen. Thanks for making time to be on The Happy Engineer Podcast. Welcome to the show. 

[00:00:11] Kristen Kehrer: Thank you for having me. I’m super psyched to be here.

Expand to Read Full Transcript

[00:00:13] Zach White: You are absolutely one of my favorite guests I’ve ever prepared for an interview Kristen.

[00:00:19] And don’t, don’t tell any of the other guests that I said that, but it’s because we’re gonna really be able to geek out today and get into some engineering speak. And that excites me. so for those who don’t know, you, you know, they heard your bio in the preview and there’s so much that you’re involved in, in data science and machine learning and, and.

[00:00:36] Truly cool technology, which, probably excites everybody who listens to this show. And so I wanna actually ask you about a little project that I saw a little side gig that you were doing. when I was up going through some LinkedIn posts of yours in preparation for our chat today of the school bus warning system.

[00:00:56] and I just thought, what a fun thing, if would you tell us about your school bus warning system project, and actually maybe geek out a little bit? Like how did you build it? What’s going on there. And maybe we’ll use that as a launchpad into some of the really cool stuff that you’re involved with at comet and machine learning.

[00:01:11] Zach White: But tell us about the bus. Yeah, 

[00:01:13] Kristen Kehrer: sure. So actually it did come about, cause I was looking for an opportunity to use the comic platform, which, makes your work that you’re running reproducible. I was looking for a use case for that and the school bus happens to pass down our street and because we don’t have.

[00:01:34] sidewalks. It has to turn around and come back, which means that I have a couple minute leeway, between when the bus passes my house and when my daughter actually has to get on the bus. And luckily it’s also at the end of our driveway that she catches the bus. So it’s sort of a. niche scenario that not everybody has, but I, built a computer vision model in Python that, uh, uses object detection.

[00:02:02] It was Yolo V five. It’s going to catch the bus going by my house and then send me a text that just says the bus passed dub and then I’m able to tell my daughter to go put on her shoes, and go catch the bus. And so it still has some false positives. I have tried using a simple classifier algorithm.

[00:02:24] actually not working any better. my hope is to really be able to, add my neighbor’s phone numbers to it for the fall, so that it’s, actually benefiting the neighborhood, but that’s where I’m at right 

[00:02:37] Zach White: now. I love this. And so false positives, what what’s driving the false positives.

[00:02:43] Where do those come from? 

[00:02:45] uh, Yolo V five, actually. That’s just one of the instances that is known for that algorithm is that it does have issues with false positives. a lot of the data I have manually created myself. I took videos of the bus passing by my house. And wrote a little script to, parse that into a ton of photos that I then manually annotated myself.

[00:03:09] Kristen Kehrer: So I am not working with, tens of thousands of photos. I am working with a, limited data set. So that is, of the challenges. And actually, that’s why I. Tried to move towards another algorithm was after, learning about the false positive issues that it has. I was like, okay, well, try another algorithm and, and see if that’s performing better.

[00:03:33] Um, but it’s also the summertime now, so I’m not able to test it as regularly. Sure. So unfortunately it’s a, yeah, it’s slow moving at the moment due to. 

[00:03:45] Zach White: School schedule issues. That’s all right. I just love this because engineers at heart. Builders creators. we just love to, use the world around us in fun and creative ways.

[00:03:58] And whether it’s, the Emmys who are gearheads and wrenching on old cars in the garage, or, our software folks who are riding code to detect buses as it drives by your house. That’s something that I see with all our clients and it’s such a passion area. So I’m curious. Kristen for you.

[00:04:15] where did your passion for all of this world of data and machine learning begin, take us to like what lit the flame for you? So my parents wanted me to study math, just because they assumed that there’d be a job market for people who understood math and I’m first generation college.

[00:04:37] Kristen Kehrer: And, I come from a blue collar family. I. And I didn’t do particularly well in high school either because I didn’t always show up. and so when I got to 

[00:04:45] Zach White: college, like literally didn’t show up, like you just didn’t go to school or do 

[00:04:49] Kristen Kehrer: you, I had a lot of classes. Yeah. The principal would be calling my mom again, saying like, Hey.know, my mom dreaded, going to parent teacher conferences, the teachers, the math teacher would come up and be like, are you happy with your daughter’s great. She doesn’t come to class. 

[00:05:04] Zach White: what, um, would you be willing to tell us what was your favorite hooky activity? If you weren’t in class, what were you doing?

[00:05:12] there was actually a McDonald’s within walking distance so, you know, it was either that or going to Cumberland farms and getting a nest cafe and a bag of, uh, salt and vinegar chip. That 

[00:05:26] Zach White: was, I mean, I, I don’t blame you. It sounds amazing. Yeah. If any kids are listening to this, go to school, it’s very important, but, um, yeah.

[00:05:34] Well, you 

[00:05:35] Kristen Kehrer: can’t get away with it the way that I did now, if I did the way that I did in high school now, I probably wouldn’t have gotten into. college that I did, it is a different world. Now this was, you know, more than 20 years ago that we’re talking about times 

[00:05:49] Zach White: have changed.

[00:05:49] That’s true 

[00:05:50] Kristen Kehrer: times have absolutely changed, but, um, To keep up because of that, I went to like seven hours of tutoring a week. I don’t know if it was just that, like, I no longer had extracurricular activities, but I did become obsessed. And, even just like my notebooks were meticulously written I took pride in, doing this and I, and I did finish in three years.

[00:06:14] I finished my degree in three years. Because once I caught up, there was no stopping me. I just had this like incredible foundation and, was able to finish my degree early with a 3.8 GPA in my major. 

[00:06:28] Zach White: this was the mathematics degree that you’re talking about.

[00:06:31] Kristen Kehrer: Yeah, my university degree, I finished for the 3.8 in my major and all around, magnet, CU Ladi. and that was able to get me, uh, teaching assistantship for grad school, which I ended up not. Finishing my first time, you know, when I went back and got a master’s degree in statistics and actually completed it, that was my second attempt.

[00:06:52] but yeah, no, I, I think that was the start something at some point clicked that made me want to do better and I’ve always been very obsessed and fascinated by. Trying to optimize my life. you know, I found myself in a dead end job with the, with the bachelor’s in math. And so then it became well, what’s paying well, where’s the demand.

[00:07:13] and that’s how I found myself in a master’s degree in statistics. And then even after that, I was a job hopper I’d stay someplace for two years and then it was like, What skills do I need to acquire that are going to get me to the next level, what job is going to teach me those skills? Where can I get my next, $12,000 pay, bump through a jump.

[00:07:36] and you know, I I’m like obsessive about it. I thoroughly enjoy that type of stuff and it translates into my life too. Like when I come into a little bit of money, I’m like, what happens if I throw towards the principle of my mortgage versus doing an inflation index bond versus something else.

[00:07:52] Kristen Kehrer: And I will just, do these analysis. And my, career itself has been an analysis. 

[00:07:58] it’s such a unique background, obviously math and statistics is a. Solid foundation for the work that you do today. And it makes sense, but not everybody who comes from that background ends up in code and development and understanding what you do around, especially machine learning and the trends in data science.

[00:08:17] Zach White: And so where did that piece, like your ability to ride a bus algorithm in Python? Right. Where, where did that begin? 

[00:08:25] if I had to do my life all over again, I absolutely would’ve taken more coding in undergrad in grad school cuz you know, getting a master’s degree in statistics. I absolutely came out being able to model in R but otherwise I, I couldn’t really automate processes.

[00:08:42] I really couldn’t find my way around in a way that, would’ve been beneficial in my first couple jobs and. Immediately start seeing. So my first job was in, econometric time series analysis and forecasting, and there would be areas where it would just, Engineering had built something before we didn’t have the hours for them.

[00:09:05] they didn’t have the bandwidth to change something to reflect what we actually needed it to do now. So it became like, Hey Kristen, can you build a macro to like fit into this thing? And so. very early on, I saw the limitations that I had myself, if I could not program. And then in my second job was where I learned how to, use sequel inquiry, a database, which again, I was seeing like, okay, when you’re working somewhere, the data that you need lives in a database, you have to be able to query for that data and just, the type of analyses that were actually going to add value to the business. I just very early on realized that for me to, continue to earn more money and to scale my career, I was going to have to be able to code to continue that growth. And I think that that’s probably like a really controversial thing to say.

[00:10:02] I think that there’s, you know, a lot of people that are very, they like their tools that they’re currently using. They’re maybe not coding and, you can absolutely have a full career where you don’t learn to code. But for me, I felt like the ticket was. Learning to code.

[00:10:21] Kristen Kehrer: And actually too, my husband is a software engineer. And so, third job , we’re both working for the same company and I could very clearly see how, the software engineers they’re building software. It adds to top line revenue. And so they were treated so much differently than myself in an analytics function.

[00:10:44] That’s a support function that sure. You might be saving the company money. Very rarely. Are you, revenue positive. I just understood that relationship and how coding was going to allow me to do things that were going to make me more valuable to the company. 

[00:11:01] Zach White: It’s a really interesting perspective.

[00:11:02] And it comes up a lot with my clients who are in your world. It’s like, do I wanna be part of, what’s seen as a cost center in the company or really working. And what’s seen as a profit center for the company and mm-hmm we need both. And so there is obviously an important debate to be had. how do you create a really amazing career on both sides?

[00:11:23] But I think what you’re saying is super important for folks to realize and think about their career strategies and how do they get experience or capable to deliver on both ends so take us into then today, you know, What, what do you see as the state of industry and the, you know, as a, a developer advocate is your official, LinkedIn title and you, you do a lot of things right now.

[00:11:47] Mm-hmm but can you just tell us because of your visibility to what’s happening? With the tools with the, stack and everything on the tech side, but also how companies are approaching this and some of the trends you know, what is Kristen’s perspective on, the state of industry when it comes to trends and data and all of this?

[00:12:06] I know it’s a big question, but what, what are the most important things you’re seeing happen? 

[00:12:11] I think it really depends on what industry you’re in, what the size of the company is. The maturity of the company, is going to have very different expectations on what they’re looking for from a data scientist.

[00:12:25] Kristen Kehrer: Right. Because we all know that if you’re working for a startup, you’re going to be, Wearing a ton of hats and you might be making recommendations on the type of software that you wanna be using. And then in that case, you get to sort of bring your preferences with you, and there’s also more opportunities for growth.

[00:12:41] And then on the other side, you have. companies that are older and more established they sort of already have their processes. the role that you’re going to work is maybe going to be of smaller scope because everyone has their lane. and you’re going to change.

[00:12:58] To be in the, processes that they’ve already sort of defined rather than defining them yourself, and I’m like super grateful that I had the opportunity to work at those larger companies. But as time has gone on, I’ve realized that for me really, I belong at smaller companies that are more flexible that are going to cuz as a creative person, I just, have to be able to, You know, think through strategy and, create and, and help to define those things rather than to fit in.

[00:13:31] Yeah. a box that was already created. 

[00:13:33] Zach White: Totally. Well, maybe if we take those two categories, so we’ll take the kind of startup world, small organization, just getting into a scaling mode. Perhaps they’ve got some fit product market fit, and they’re looking to scale versus the. Fortune 500 large organization.

[00:13:50] If we just took those two for a second, I’ll give you a challenge. And I know it’s a challenge because I’m talking about myself here as a mechanical engineer who doesn’t live in your world. If you were tasked to we’ll start with the startup side. So let’s say I’m the founder or the technical co-founder of a startup.

[00:14:08] And I really have a product, mechanical type of lens. And you were coming in and making a case. Around really what we need to be focused on as we scale the organization, when it comes to data and all of what’s related there, what would you think I need to understand about what my company must get right.

[00:14:29] To be successful in the next with what you know now. So starting on that small startup side of. like, I don’t know data sets, 

[00:14:37] Kristen Kehrer: but there, yeah, you wanna be in the like move fast and break shit. And to allow yourself to do that, you have to have the correct testing capabilities and I’m a big fan of.

[00:14:50] Ad testing. even in like a, sometimes non rigorous sense, cuz especially with these smaller companies, we don’t always have enough data to have the sample size to say something. is statistically significant. Um, the smaller organizations, when you have somebody who maybe owns, social media, Strategy or whatever.

[00:15:13] And, and you wanna test something. There can still be that pushback of like, well, you know, I defined this process and this is sort of my area, it’s easier to, I think, step on toes in a smaller organization because you wear so many hats and so many things are, cross-functional and you’re all working together.

[00:15:32] before anything else before the ML comes cause at some point, you’re gonna wanna be optimizing for your ad spend. You’re gonna wanna be looking at retention. that’s when you’re at the point where you’re, actually running ads and you have a bunch of customers and you’re thinking about how you’re retaining.

[00:15:49] which I think is, maybe a little bit further along in the process, but it, in the beginning, like, testing and being nimble is everything. and, thinking through the couple KPIs that are actually something that you can influence and move and, tracking those and setting that up in a way that, isn’t going to, take somebody’s bandwidth.

[00:16:12] Kristen Kehrer: for multiple days a week to, to create some report that has just like more than you need, but yeah, I’d, I’d say, staying nimble and, and also thinking about how you’re collecting your data in a way that’s going to allow you to scale and have the things that you want later on mm-hmm 

[00:16:27] Zach White: this is interesting.

[00:16:29] course the engineering leaders who listen to The Happy Engineer Podcast and whoever’s with us right now, listening is probably oh yeah, you know, check, check engineers, get it. But I hadn’t met a lot of CEOs who don’t come from that background who would necessarily associate machine learning and data science as a critical function of the business during a scale up phase.

[00:16:52] Zach White: To rapidly accelerate, great decision making. And you mentioned employee retention, like who who’s out there thinking about how to leverage machine learning into retention. Like that’s such an interesting connection to me. And so I don’t know, could you just help us put a little bit of perspective around what does it look like to build a learning organization?

[00:17:15] with data as the foundation, like, is that something that you just have to shift your mindset and go after? Or is there like what what’s missing? Cause I don’t see it that often. I talk to all 

[00:17:26] Kristen Kehrer: kinds of it’s bringing in the right person. You need that, real innovative.

[00:17:32] Director level or senior manager who understands that if they come in and they create. Some real processes and really, and of course there’s, data engineering and we need to have the data infrastructure and we need to have buy-in from yeah. the right people to invest in that area so that the data exists.

[00:17:50] But the side that I normally sit on, the, the, the problems that we see are that, you know, marketing wants to be able to do what they. To do. Um, they come up with a great idea. They wanna be able to put ad spend behind that. and a lot of times, these companies will have a 20 million budget for ads.

[00:18:10] Sure. That’s never been optimized. and so, right. There’s this friction between like, Hey, we want to, have, some. Say on, on what it is you’re doing. And they’re like, but I want to do my job. And that, follows into the testing as well. They’ll say, you know, Hey, can we test these things?

[00:18:28] And it’s like, well, Not in a meaningful way because of X, Y, and Z, and it doesn’t have enough sample and it would take us three months to get to a place where we’d actually be able to read out on that. That should be a strategic decision, whate all these places where you can butt heads with your stakeholders.

[00:18:45] And, and so you need that senior manager level person, who’s gonna come in and be like, okay, now we’re inviting marketing. To our stand up so that they have visibility into what we’re doing. We’re gonna create a process around how we’re doing testing and every meeting’s gonna look the same. They’re gonna call out, asking for a test around something.

[00:19:05] We’re gonna set up a meeting. We’re gonna go through this thing. And it’s, you know, just so that when you say no to a test, it’s not because I. Mad at you or we have a personal week. It’s because it’s like very clearly documented in our process. Right. I think that the trust between the data science and stakeholders is one of the biggest things that keeps The data, analytics, maturity from moving forward. there’s this constant, difficulty in communication. And I think that once you get through that and you get, buy-in like, there’s just nothing totally in your way. That’s stopping you. You’ve, you know, at that point, once you’ve got trust and you say, Hey, like we need to, we wanna take the.

[00:19:49] Four months and really focus on how we’re going to think about retention campaigns. And we’re gonna do this through building a machine learning model, then you get that, buy-in better. I can’t tell you how many models I’ve built that, uh, never went past the POC phase and it’s been cool cuz at least, I can put on my resume like, Hey I did this.

[00:20:11] and I actually did it, but you know, uh, few things have gotten to production. This, the relationship is the hurdle. Oftentimes. If you’ve got the data, right. Like I don’t, I’m sure there’s people sitting there listening, being like the data and it’s like, yeah, I, 

[00:20:24] Zach White: you have the data. Yeah. But you know, as a, as a coach, what I hear behind this as well, and it’s like, there’s so many data scientists or software, engineers who have a passion for ML AI type of work that.

[00:20:38] Feel uncertain or confused about what type of opportunities they can create for their career. And this, to me, just, it’s an entire space where. I mean, I don’t know how many thousands of companies in this small to medium startup kind of world looking to scale and grow who have none of what you just described really in place who could benefit tremendously from optimizing their 20 million ad spend, but don’t, and if you can go in and make that case and be that director, you could go from.

[00:21:10] senior engineer at some big company to making a massive impact in a, $50 million small organization. If, if that’s your passion, get more open and creative about what kind of roles, you know, you could go make the job of your dreams in a company doing work that you’re passionate about.

[00:21:29] Zach White: Anyway, I just am super encouraged by this. I hope people are hearing what you’re saying, Kristen, and maybe the wheels are turning. I don’t have. Right. Well, cause once you 

[00:21:36] Kristen Kehrer: see it, there’s all this low hanging fruit, right? Yeah. You can walk into an organization once you’ve, you know, focused on retention and acquisition, and ad spend for a while.

[00:21:46] It’s like you can walk into any eCommerce company and oh yeah. And if those things aren’t done yet, Like, rinse and repeat, but to the people that you’re just proposing it to, they’re like, wow. Yeah. 

[00:21:58] Zach White: exactly, exactly. Okay. So let’s flip to the coin then if we’re talking, I’m a data scientist or, working in, in a fortune 500 organization in this world, where do you see the biggest leverage right now for people who are on that side of the coin?

[00:22:14] Kristen Kehrer: My experience over in the larger companies has historically been that it’s difficult to find the growth that I was, really desiring for myself. that’s always been a sticking point for me. So, you know, for me personally, I’d go and, pick up some amazing skills.

[00:22:38] And after two years I’d be like, okay, maybe I’ve gotten what I’ve needed to get here. And I. Start looking for other opportunities. So there is, you know, huge opportunity to learn and advance your career. But for me personally, I’d find myself in these companies and, I’d just be looking at the org chart and I’m like, I’m not seeing how the like, upward progression is gonna come.

[00:23:05] And so, you know, as, time goes on, I do have less to say about those larger organizations, because it really hasn’t been, where I feel like I fit. I also feel like, the smaller companies are more progressive and so I’ve always felt like I’m able to be my more authentic self. so for like a number of reasons.

[00:23:23] Yeah. I feel like I Don. Have it to speak on these larger organizations anymore? 

[00:23:31] Zach White: Yeah. Well, I appreciate your candor on that. And I hope if, if the engineering leader listening happens to be a director or VP in one of these fortune 500 companies, just take that perspective to heart. How are you creating a more clear pathway for people in this world who are passionate about machine learning and AI and data science, and how to leverage it into value in your company?

[00:23:52] Like if they don’t wanna be directors of a hundred people, they wanna. Drive business results through this work, what’s their option. So it’s a problem that needs to be solved. Mm-hmm Kristen, what are you most excited about when you look at how the technology is developing? you go on LinkedIn and every other data science person has some sort of, ML, AI.

[00:24:14] Thing listed in their title and wants to be associated to this as a trend. But, you’re really involved in a huge voice in this world. with the role you have now, what are you seeing that looks like truly exciting in terms of trends in the technology? 

[00:24:30] Kristen Kehrer: I feel like what’s most exciting to me is that we now have the tools to actually do our job, right?

[00:24:35] 10 years ago I’d start a new job and they’d be like, here’s your laptop with two gigs of Ram now go and build me a customer segmentation you couldn’t even, you couldn’t even, the model would run for a couple hours and then it would abort and the stakeholders are wondering where your work is.

[00:24:54] all the things that I used to do manually that took up so much time are all of a sudden, like, that’s, that’s the most fascinating thing to me is, seeing the XG. And the new, algorithms come out and then, getting to a place where like, oh, okay, now we have a way to look at feature importance for these algorithms.

[00:25:18] . Which tools are really changing the game in this space. When you say we finally have the tools to do the job. It’s just about how much easier all these things are becoming.

[00:25:27] Kristen Kehrer: Right? So like, I can now pop up a dashboard using our shiny in an afternoon, like leveraging some old code or whatever, but the code has become so intuitive and things aren’t as manual. Right. Right. Like before, if you wanted to do text analysis, you might be stemming those words on your own. Like. Like we have libraries now to do things that we used to have to do manually, and there’d be, analysts all over the world, working manually by themselves.

[00:25:58] And now we’re all connected by, this open source world and using these same libraries and we are able to leverage the work that people have done on the other side of the world and for free, you know, I think like that’s really beautiful and it’s also just really streamlined the work in a way that I can just think about how long.

[00:26:20] analyses used to take me 10 years ago and, and look at, one function in a package might get you your whole ed with, the histograms for depending on the data type or whatever. It’s gonna create a histogram versus some other type of chart or give you the summary statistics with one line.

[00:26:38] Whereas before you’d be. Writing it out for each variable. Yeah. Okay. Now let me get, you know, or you’re writing a loop yourself or something to get it. that’s what I think has been like the most interesting to watch over the last 12 super 

[00:26:51] Zach White: cool engineers. I think back to my design engineering days as a mechanical guy you know, creating, uh, sheet metal parts for washers and dryers and whatnot.

[00:27:01] There was nothing even close to what you’re describing, where, you know, you could just tap into like a library of geometry for a CAD model from anywhere, anytime, and exactly, it was always custom and one off. And actually it would be curious. So if any engineering leaders in the design space have cool updates for me, that those exist in, in the mechanical world, I’d love to know, but maybe there’s a lesson to be learned here about how.

[00:27:26] This open source model is creating such incredible value across the whole industry for data science. You know, how can we. Some of that into other areas of engineering without losing what we need in terms of things that are proprietary or IP, et cetera. And mm-hmm, , I am curious from the, your side, Kristen, how is the dance between like what’s intellectual property, we need to protect versus what’s open sourced and we can all leverage for the benefit of everyone.

[00:27:55] How do you define like where that line ought to be? 

[00:27:58] Kristen Kehrer: Honestly, I’m not good at that at all. Cause I’m the type of person I’ll go build something and I’ll go put it out there to share it. You know, somebody will be like, oh, you should have done something with that. And I’m like, well, I dunno. 

[00:28:12] Zach White: I think that’s, that’s kind of cool actually.

[00:28:13] So, so really in some ways each individual. Leader in this space needs to make that decision for themselves is what I’m hearing. So if you, if you wanna, yeah. 

[00:28:22] Kristen Kehrer: Or I think that maybe there needs to be like more education around exactly where you might start to think of something as IP versus yeah, Or actually, certainly you just get rid of that question cuz it’s, it’s just really not in my domain.

[00:28:39] Zach White: yeah. Well, and that’s true. And, or really important asterisk. Each engineering leader, listening needs to understand their company’s stance on this and make sure that they’re really intentional to honor that. I love where the space has gone and the fact that it’s accelerating so quickly and I do have one. Very, you know, mechanical engineer ignorance question for you and for anybody who’s maybe been losing some of the conversation cuz they don’t know the acronyms or they’re not in this world. Can you describe what you see as the distinction between what falls into the machine learning?

[00:29:13] Side of development and what would be considered artificial intelligence. And is there a clear line and difference or for you, is there just a lot of overlap and we use the terms kind of in a muddied way. 

[00:29:25] you know, in like 20 10, 20 11, I was building, um, neural net models to forecast hourly electric load.

[00:29:34] Kristen Kehrer: And I’ve been on a couple podcasts where people have like referred to that as AI. And like I cringe, right. Because I was, very manually. So I. Train that model over the weekend. And then each day I would produce a new forecast, based on the updated weather data. And then, the following weekend I’d retrain again, but so my hands were touching it all the time and the decisions that they were making based on that output on, this was used to input, um, sort of where they should direct, electricity to keep people up during, during peak load times.

[00:30:13] that’s not artificial intelligence, right? because it was not making any decisions intelligently at all. There are people who are going to say that at the end of a couple of logic statements, if a decision is made. And executed that is technically artificial intelligence.

[00:30:32] Kristen Kehrer: And they’ll say the same things for, a logistic regression model going back to, um,were talking about retention model. Somebody builds a, logistic regression model or uses XG boost to create a retention model and then says like, okay, well, if your score is below this, let’s send you.

[00:30:52] Communication. that’s technically by definition. the machine is making a, a, decision about what to do without human aid. somebody might call that artificial intelligence. That’s just not interesting enough for me. I’m not exactly sure where yeah, yeah. Where the line is for me, but I would feel very uncomfortable calling any of the work that I’ve done, artificial intelligence.

[00:31:20] when I see people doing interesting things with N NLP, computer vision, things that truly require deep learning, I’d happily classify as artificial intelligence. And then of course, there’s artificial intelligence and then there’s like the piece of.

[00:31:39] computer vision, NLP, deep learning in that bubble, but artificial intelligence is also, much more than that. I’m not sure if that’s helpful, but that’s sort of yeah. How I’ve been currently thinking about that space, 

[00:31:55] what’s at the very cutting edge of the artificial intelligence world in your point.

[00:32:00] Kristen Kehrer: So I was in Germany, like two weeks ago and I saw a robotic lawn mower they just have them there mowing their lawns. and it was the first time I’d seen one. So of course I like stopped to see, like, what’s the name of the company I’m gonna look them up. Right. Cause it wasn’t, I a lot.

[00:32:21] the corporate office is probably like 20 minutes from my house and I’m familiar with Roombas, but this was the first time that I had ever seen an automatic vacuum. and it was by some company that I had never heard of before. You know, when I see those types of, machinery, like actually working in real life, I’m like, wow, that is awesome.

[00:32:42] It’s harder for me to, classify, think about, how I’m classifying things when it is, in the machine learning 

[00:32:53] Zach White: realm. Yeah. Yeah. I I’m with you. there’s so many little pieces that are, it’s like right at that tipping point of like, Ooh, this is gonna get super interesting.

[00:33:04] Really soon, just kind like, kind of buckle up the next. Decade’s gonna be exciting so well, real quick, Kristen, this is, The Happy Engineer Podcast, and I always love to talk about lifestyle things, and we talk about lifestyle engineering at OACO a lot with our clients and who I coach. And you shared with me before we hit record today, kind of your own perspective about that.

[00:33:27] And I just was wondering if you’d share a little bit. You know how you’ve balanced and how you think about career and life and how that plays out for you today. And maybe what you’ve learned over the years, the hard way, et cetera. But would you be willing to share some of that, that you told me earlier?

[00:33:47] Yeah, 

[00:33:47] Kristen Kehrer: so, I mean, I absolutely have sort of meticulously engineered my whole life. you know, I found myself in a dead end job after finishing my bachelor’s degree. And so going to a master’s degree was very much, what is the demand for the type of roles that, what.

[00:34:07] Type of master’s degree. Could I get after getting a ma a bachelor’s degree in mathematics and what is going to be the demand for those roles? How much is that going to pay? What is the long term, benefit or, potential ROI and getting that master’s sure. Oh man. When people don’t think about the ROI before going for a masters, that one of thoughts, but yeah, you know, it.

[00:34:34] Getting into a co a company right after my master’s in statistics, it was very much like, what am I learning here? What is the value that I’m getting? it sounds rough, but like, now that pensions, aren’t a thing, you know, loyalty is, I, I just really have always worried about myself first and a lot of that is like, in what ways do I think I’m gonna be growing here in the next six months?

[00:34:58] What is it that I think I’m going to be able to learn? That’s going to be relevant to me. And if I can’t, see much of that, it’s like, okay, well, what does the job mark look like now? you know, I’ve hopped industries three times. I’m like a very, I’m very big into trying a new industry.

[00:35:15] And then, you know, if I go to try and get into a fourth industry and somebody says, Hey, what makes you think that you’re going to be able to learn this industry and be successful here? It’s like, well, I’ve successfully hopped industries already done it before. Yeah. Yep, exactly. I’ve done it before.

[00:35:30] and always just sort of looking for like, How much is my next, role? what’s that salary going to look like? I love interviewing, I absolutely love meeting new people and, chatting about the role that they have available. Seeing if there’s a vibe at this point in my career, it’s.

[00:35:48] Kristen Kehrer: So much about the vibe because you are spending 40 hours a, a week with these people. it is so incredibly important that you enjoy their company, I’ve always thought so heavily. I spend a, real significant amount of time thinking about my career, thinking about how I wanna position myself, thinking about how I’m going to get to where I want to be.

[00:36:10] and thinking about salary. 

[00:36:13] Zach White: I mean, I love this and just that proactive be in the driver’s seat of your life and career approach. And a lot of engineering leaders, I talk to, they go 3, 5, 10 plus years, just kind of drifting and enjoying what they have and not to say that it’s bad or there’s no judgment of that, but kind of pull their head up after a decade of being on one track and wondering, whoa, like a little disoriented, where do I go from here?

[00:36:39] And what have I been doing? things change quickly in, in our, our world as engineering leaders. So I think that’s awesome, Kristen, that you’re so proactive in that. 

[00:36:46] Kristen Kehrer: uh, well, there are people that are lucky, right? Like there are people who are going to be in a job and they’re going to get an amazing boss that is going to advocate for them to get promotions or, you know, by some stroke of luck they’re going to happen to be in the right, org in the right area, under the right person.

[00:37:04] Right. Mm-hmm, , there’s just a lot of factors there, but a lot of. Aren’t going to be in that position. And so then it really does become, you know, Working for yourself so I can see how somebody can spend 10 years and, be really lucky and, and not think of it a whole lot. And it’s because, circumstances have just been favorable for them.

[00:37:24] And that’s wonderful. good for them. 

[00:37:26] Zach White: yeah. Yeah. And, but I would just still echo, even if you’re in that situation and, and you got the great boss and everything’s going. It only takes one little thing to flip, right? Your boss leaves, you end up with a new boss and it’s, it’s a horrible situation suddenly, right?

[00:37:41] Mm-hmm, , it’s the old adage that people join companies and leave bosses. And so it’s a reminder though, even if you’re in a good place, don’t fall into that comfort zone of not doing the work to understand what’s happening in your industry. Are you learning and growing? What other opportunities are out there?

[00:37:58] Even if you have no intention of taking them, make sure that at any point, if your situation did change. that you’re prepared to be able to respond, without feeling like you’re way behind. So mm-hmm, , I think that’s important. Kristen, this has been awesome. And I want you to make sure and tell people if they wanna, connect with you and also the amazing work and the tools and the platforms that come brings to the table and everything with, with your role there, where can people.

[00:38:24] Reach out and connect if they wanna learn more or just maybe get exposed to some of the amazing work and content that you’re creating these days. 

[00:38:32] So I spend a ton of time on LinkedIn. Um, so that’s always the number one place to go look me up. I do have a blog it’s data moves and so I’m actively, still posting there.

[00:38:46] Kristen Kehrer: I’m currently breaking my computer vision model. Project into little pieces and writing blog articles about that. so there, but definitely I’d go to LinkedIn first. 

[00:38:57] Zach White: Awesome. So we’ll make sure your LinkedIn profiles in the show notes. So people can find that they know where it is and The Happy Engineer and Kristen, just to land the plane here, I.

[00:39:12] Get curious with our guests about how important asking great questions is, because whether it’s great engineering or great coaching, we know that questions lead. Answers follow. And if we want better answers in our life, let’s ask better questions. And right at the end there, we touched on a, a cool thing.

[00:39:32] You’re probably in the minority of engineering leaders who likes to interview and for sure as a leader in this space with data science and machine learning, et cetera, I’m curious for that engineering leader, who’s been listening to our conversation and wants to advance and develop in that way. Would you encourage them in the questions they should be asking?

[00:39:56] I think to be a happy engineer, , uh, you really need to know what it is that you’re getting into. And I think that we don’t talk enough about, um, you know, a lot of times when people are talking about the questions that they’re asking. Particularly in interviews, it’s more, behavioral, or if you ask this you’ll, you know, sound smart and whatever.

[00:40:20] Kristen Kehrer: But I, I think that really, what we need to be asking is like, what’s the state of data governance in your organization. What is the state of self service tools? How siloed is your data? What type of infrastructure do you have? Are you able to, if somebody wanted to create a, uh, customer journey from the first time they visit the site until, present day, like are the.

[00:40:47] Systems with the data, talking to each other in a way that I’m able to get at that, or is it like this huge manual lift where I need to go talk to some other group to be able to get at all the data that I need to empower me to do my role. And so, it’s very easy to be an unhappy engineer.

[00:41:05] If you. Walk into a organization without that information and you, and you don’t, you know, you don’t know what you’re getting. I think that a lot of times people don’t ask the questions and on day one, they’re like completely blindsided by the real state of the state of the data that they’re going to be dealing with.

[00:41:26] Zach White: I, I love this. I hope people are paying close attention because you’re so right. You know, and I’m guilty of this in some ways, too Christian, where we focus on interview strategy and how to, interview well on those other buckets, asking great questions. Everybody wants to know what to ask around culture and fit and all these things, but to get genuinely.

[00:41:48] Deep and curious about, your domain, the state of the data. And I would just challenge every engineering leader, listening. If that’s not your world, maybe you’re a civil engineer or you’re a mechanical engineer, maybe you’ve, you’ve gotten out of engineering altogether. You’re working and just running the business.

[00:42:02] But. If you’re making a transition or you’re considering a new role for your career, what is the equivalent of that question for your space? What do you need to know about the work and what’s available and the tools and the data is that’s awesome. Kristen. I love it so much. Thank you. And thanks again for making time to be with us.

[00:42:21] So you have lots going on and, just continue to do that amazing work and content. And let me know how the bus algorithm goes this fall. Maybe I can get on your, text list to get the bus warnings. See if I’m not on your street, but it would be kind of cool to get that. thanks again, Kristen. This has been.