On this episode, Susan Ottmann talks to Tony Orzechowski from Abbott Diagnostics R&D Data Analytics organization about how he got started in data analytics and why it’s important for engineers today. 

Tony Orzechowski is is the director of the Abbott Diagnostics R&D Data Analytics organization and an instructor in the Master of Engineering Data Analytics program at UW-Madison. He has a distinguished career with over 35 years of increasing management responsibility. He is responsible for both ongoing support of product launch and digital innovation with over 100 statisticians, data scientists, business analysts, quality engineers, and data management personnel. He received a mechanical engineering undergraduate and a manufacturing systems engineering master’s degree from UW-Madison. Tony is a certified Six Sigma Black Belt, an adjunct lecturer at Northwestern University, and is launching a new non-credit course Data for Technical Leaders at UW-Madison.

Subscribe to Podcast

Justin Kyle Bush  0:03   

Welcome to Epod, a podcast from the UW-Madison’s College of Engineering Office of Interdisciplinary Professional Programs. These podcasts are focused on big ideas in engineering and the people behind them. My name is Justin Kyle Bush, and I’m your host. 

On today’s episode, Susan Ottmann talks to Tony Orzechowski, from Abbott Diagnostics, R&D data analytics organization, about how he got started in data analytics and why it’s important for engineers today. Take it away, Susan.  

Susan Ottmann  0:43   

Welcome, Tony. 

Tony Orzechowski  0:46   

Hey, thank you, Susan. Thanks so much. And first, I just want to thank you for allowing me to share some time with you on this. This is exciting. I love this topic. And I think we’ll have a good time here. So, I think you want to talk a little bit about our history in how I got started with analytics, right?  

Susan Ottmann  1:06 

Yes. Before we start, how about just tell me a little of your story?  

Tony Orzechowski  1:09  

Yeah, absolutely. So, you know, I’m going to take you in a way-back machine here. I’m a UW Madison graduate, mechanical engineering, about 37 or so years back, and in my first year, I was assigned into some Process Automation work that was just incredibly challenging, a wonderful project. But you know, that was a, I think, a key part of my future, because it really, you know, the fact that we had to automate and hold such high levels of quality on processes that traditionally had been very manually oriented. And so therefore, adjustments can be made by the technician to adjust for perhaps, changes in tensile strength of materials, or other incoming raw material variations, environmental conditions. And at that time, you know, I really didn’t have a background in analytics.

Really, in the early 80s, that wasn’t prevalent for engineers, they have that background. And with the challenges of this particular project, I was lucky enough to have a manager who was an incredible mentor for me, throughout my career, who pointed me in the direction and said, “You know, I’ve heard of something called the design of experiments, you know, Tony, take a look at that, because that maybe that could help us in this very challenging project that we have to understand, you know, how we can drive better quality through variance reduction, among others.” That, for me, was a monumental key point in my career. I didn’t realize at that moment, but I look back now. That’s how I got myself into analytics 37 years ago, was going to my first courses on design experiments. You know, under George Fox and Conrad Fund, Swarm Biscard, all from the University of Wisconsin. I was located down in Racine, Wisconsin, at Modine Manufacturing. And that was a point where, as I learned it, things just opened up for me. I began understanding how to use analytics to characterize the factors that actually drove the variation in my outcome responses. It gave me a broader view of all this, it began to teach me about the idea of pattern recognition, signals, and noise.

And all of that really was what allowed me to be successful in that automation project that we had going and really opened the door to me becoming a master black belt over time, returning to Madison for my Master’s in Manufacturing Systems Engineering. And then my career had added for the last 27 years, leading an analytics organization. But it really all started with an engineering challenge that couldn’t be solved with engineering alone. It really needed to be complemented with not just mechanistic understanding, but also empirical understanding. And then empirical understanding was all around analytics. So, it’s kind of an exciting start, and makes me think back of some exciting times in my career. 

Susan Ottmann  3:56   

Yes, definitely fascinating. Let’s dive a bit deeper and talk about data analytics and why it is important for engineers today.  

Tony Orzechowski  4:04   

Sure, sure. Well, I think of three things when I think of your question here. One is you’ll just for the traditional engineer coming in, who needs to look at data. And you know, very traditionally engineers, you know, we’re characterizing our process. We’re trying to figure out, you know, what are the key influential factors that drive my outcomes? We’re looking at times to optimize our outcomes, you know, to drive to minimum variation or maximum yield, among other things. Often, I’m also looking to understand the capability of my product or process to meet output requirements. And then finally, I’m also looking to run and design experiments around investigational, when things aren’t going quite right, or I need to improve further, how I do that. All of these are, you know, data-driven type of activities, that when I compliment again, my engineering background is supplemented with data analytics, it’s a really powerful combination, because I can understand very quickly how to separate signal from noise, understand how to make decisions better, more rapidly and do it in a very effective way. So, it’s not as if analytics would replace it, but it supplements it. That’s path number one.

Path number two, which is becoming more prevalent today, is where analytics becomes the product itself. I’ll give you one example from my current industry, which is in healthcare and diagnostic testing. Is where we’re looking at taking multiple sources of information: demographic information about the patient, information about the results, test results, we produce on your blood or other fluids that are tested, and bringing this together, using machine learning and other methods to actually produce a probabilistic outcome of your potential for future healthcare incidences. If it might be heart disease, it might be cancer, or others. You know, blending these multiple data sources together, the analytics, the algorithm itself becomes the product, becomes a medical device, okay. And that’s an exciting field, where analytics, you take your engineering or science background, and now it becomes, again, a new product versus just a support to the engineering activities.

So that’s a second path, then finally, we know our leaders, including myself, you know, really are always looking to say, how do we make our best decisions under uncertainty, you know, when there’s variation present when there is uncertainty, and our decisions are not deterministic, or probabilistic. And again, using statistics and using data analytics to extract out our best estimates of what will probably happen in the future under various outcomes. That, you know, translate that and talk about risk and quantify the level of risk and probability is extremely powerful. So those three pathways for engineering are I think where analytics really complements it very well. 

Susan Ottmann  6:57   

Over the last year, you and I have discussed different areas of data analytics. When you think of data through the lens of an engineer, what key areas must they have proficiency?   

Tony Orzechowski  7:08   

Yeah, I think it links back to what we were just speaking of. It’s signal and noise. If that’s pattern recognition, if I’m doing vision systems, I’m doing acoustics, or if I’m looking for anomalies and idea, it’s really all about recognizing patterns and recognizing relationships between input and output factors. This is what’s going to allow me to be able to detect where something’s different. Again, if it’s facial recognition, again, if it’s looking for defects in my chip manufacturing, whatever it might be. And then if I go one level deeper, if I’m going to be really good at the analytics, I also have to primarily focus on how I design for analytics, what kind of experimental work will I do, if it’s A/B testing, if it’s factorial designs, if it’s randomized experimentation. But I need, as an engineer, one of the things I’d highly recommend for any engineer, is to go into depth of understanding how to design the best experiments that can be then leveraged for analytics, either by yourself, or by an analytics support team. Which is what we have here in my company, is our engineers do both. They analyze but they also come to a group like mine for very detailed analytics. But it all starts with really designing for those analytics effectively. And I’d highly recommend that for any engineer, if they’re going down the analytics path or not.  

Susan Ottmann  8:33   

One area I’d like to explore further that I find interesting is active experimentation. Please explain the theory of active experimentation and your viewpoint on how this can accelerate design cycles.  

Tony Orzechowski  8:46   

Yeah, sure. You know, active experimentation. I think that’s the theme we’ve been talking about, and I love it. I think it’s the single most important skill an engineer can have and an analytics person, for sure, it is understanding active experimentation. And I’ll use the term design of experimentation. I look at active experimentation as a broad set of activities that start from understanding the business outcome or the system level outcome I’m looking for, taking it down, and then understanding the responses that I measured, being able to characterize it. Look at those responses over time. Look at the robustness of them. You know, how variable are they? How well does my measurement system measure those responses, right? If I can understand that, that’ll give me a sense of the capability and how far I need to go to improve. 

That foundational piece going into the design of experiments, then drives me into: How do I select the factors that I may want to do active experimentation on? How boldly can I begin to separate those levels from low to high as I try to understand their effect on the outcome when I’m looking at the response? That bold experimentation, which is really there to amplify the signal, you know, in our design experiment approaches.

And then finally, the parts I begin looking are also the level of noise that’s present in my response, and in my factors and strategies of how I reduce that type of noise. To again amplify the signal to noise ratios to be able to say, “Yes, I can tell you that here’s the key influential factors.” And if I do that and understand the concept of which factors may confound my experiment, how to randomize it correctly, all this comes together to really provide me an experiment that tells me, here’s the key influential factors. But also, I can begin building models for optimization and really tuning my end effect, most importantly. So again, sound foundational experimental methods, I think are probably the most important skill, an engineer, outside their engineering domain, that they can complement their current process with. They complement their mechanistic understanding with empirical understanding. 

Susan Ottmann  10:54   

I find it quite interesting. I did a lot of the “Oh, when I was a young engineer,” and the field has come quite far. And the idea of getting what’s the noise and understanding that noise is so important. Another area that aligns with your work is storytelling with data, please explain this concept and how you’ve applied it to projects in the past.  

Tony Orzechowski  11:17   

Sure, sure. If I start with what I believe, my own personal view on the purpose of data analytics, is really simply to obtain meaning from data, any meaning from data. And that’s the key to analytics. However, to really have an effect, that’s not enough. You have to be open to communicate that meaning to others, if it’s your colleagues, or if it’s leadership. And that’s where storytelling comes in. Because, you know, myself, and over the years, I probably under-appreciated storytelling in my career. Only recently, in the last five to 10 years, did I really become, as I moved up in my career, to understand the power and the importance of storytelling. Because at the end of the day, we want to believe that somehow the data will speak for themselves. And the story will come out. And this, really, if you look how decisions are made by leaders, it’s an emotional event. It’s one that uses their experiential, their intuition. Yes, the data influences it. And that’s where, again, being able to effectively tell a story with your data that, you know, talking about “the big idea, my unique point of view, here’s what I believe my data is telling me,” but also what’s at stake. That creates emotional connection to your audience, if it’s a leader or others, because what’s at stake for them really causes an emotional event, which allows them to retain the information, allows it to resonate with them. And it really will allow you to, again, you know, to bring your compelling story forward in a way that they’ll embrace. So, you know, looking at that part of its really critical.

And I’d probably throw one other element on. You know, if I look at the leadership’s role in analytics, I look at leadership as having the key role with data analytics. You know, it’s not only just the analysis itself, what the analytics people will do, but if you look at it, all the key problems that really are important to the business are owned by leaders. If you look at who owns the resources to solve these problems, again, it’s our leaders. The planning process that will allow us to effectively plan the right experiments with the right resources and capital and expense dollars is our leadership. And for them to have a sound grounding, you know, through storytelling and through other means, to be able to understand what you’re trying to convey to them, they’re the ones that will control that your ability to be successful. So being able to speak to them in a way that will resonate with them best is very, very important to the career of anybody who’s in analytics, especially as they move up, you know, driving, you know, broader and broader types of decisions.   

Susan Ottmann  13:48 

So, we’d like to say one of the objectives of the program is to turn data into decisions, and you need to articulate that, and that’s really in that data storytelling.  

Tony Orzechowski  13:59 

Yeah, absolutely. Absolutely. I completely agree with your premise there, the decisions and the actions that will follow from those. Absolutely.  

Susan Ottmann  14:05 

Tony, students enrolling in the engineering data analytics programs are often career changers, moving from design or operations into a more full-time data role. Any advice for those looking for a career in data?  

Tony Orzechowski  14:18 

Yeah, yeah. You know, I’ll probably give you two pathways here. One is for the engineer who wants to remain an engineer, still wants to be designing product, but they want to use analytics, right? This kind of a program, you know, at Wisconsin is an incredible benefit to them because it shows them how to complement their engineering background, again, all the things we talked about earlier. If it’s algorithm development, or if it’s characterization, capability analysis, optimization, all those things are true. 

You know, the pathway that I pick, again, as I transition from an engineer to a data analytics person, and that was a transition, right? That happened. I started as an engineer, you know, complemented my engineering background with analytics. And then as I saw my passion in this area, I really focused heavily, completely on the analytics in support of engineers. And that’s that, that’s the next step for me was transitioning to an analytics support role. And, you know, the beauty of that is, you get to impact all products. You know, at least here, our group is probably on any given day, associated with 30 to 40 different products that are ongoing in our R&D organization. And that breadth of impact, I couldn’t have it, you know, if I were, if I remained as an engineer. That’s important to me, you know, everybody’s different. But that’s important, to be able to have that breadth of impact.

But I will say it does take a level of, I guess I would call professional maturity, because you have to understand that you’re stepping away from, I’ll call it the direct contact to the product as an engineer, where you’re designing the physical elements of it. And you’re moving sometimes into a support role for the engineers to be able to help them with their analytics and drive those things forward. And you’re a step further, you know, in this supply chain, you’re one step further back. Again, the breadth is awesome. But you have to understand that you are no longer what the engineer was, and you have to accept that role. And that takes some thinking. You have to really be ready for that decision.

One other part that I will say is, an engineer who moves into an analytics role is a really valuable entity because they become translators. And I’ve read some papers on this topic I so enjoyed that really talk about the fact that if I can have somebody in the analytics team – like I do today, I have both engineers and scientists who actually develop products – when they work with their analytics with the science team or the engineering team, they have a level of fluency, and credibility, because they understand the engineering, they understand the challenges that those folks are facing, they can translate the analytics into a way that that scientist or engineer can understand. And then they can actually, again, focus on the problem much more rapidly and effectively. That’s an incredibly powerful role for an engineer who wants to transition to data analytics, to think of themselves as an analytics expert, who has an engineering background and becomes a translator between those organizations. That shouldn’t be underestimated, the power of that particular service. 

Susan Ottmann  17:15   

Before we go, is there anything else you’d like to tell prospective students listening to this podcast, or others interested in learning more about engineering, data analytics, and its application in industry?  

Tony Orzechowski  17:26   

Yeah, yeah. I think as you start down the road, and you look at UW’s program, and I think UW’s program is unique because I think it hits the right elements. It is not only focused on the technical elements of machine learning, it looks at the breadth of analytics. It looks at even the use of statistical methods for descriptive analytics, it looks at storytelling, it looks at visualization. I think these are all really important elements. And that will give you, as generalist analytics, just a great overview.

Within that program, I’d highly recommend you find the area that you feel the greatest passion for. That might be, again, if it’s descriptive analytics, describing what’s happened, building dashboards, visualizing it, finding those, you know, those anomalies and saying, this is where we need to focus for the future for management or others. Or if they happen to be causal analytics to see why something happened, what were the causal factors, or if it’s optimization, right. All of those, each of those are very deep domains. And you know, becoming great at all of them is very, very difficult. I would say, find the area that you really have a passion in and go deep in it, and really become great at it. Again, you can expand and move into other areas. But again, find your passion within analytics, because it is such a broad, exciting area. And it’s not going away, it is only becoming stronger and more important as we move into the future with lower cost data storage, greater computing power. You know, our sensor technology, our ability to embed analytics in our products is growing.

With all these things happening, I think the future for an engineer with an analytics background is just outstanding. I just again would just say, focus on your passion, and that’s going to take care of itself. 

Susan Ottmann  19:16   

Well, Tony, thank you so much for your time today. I find your perspective fascinating and your energy exciting. And I believe your passion for the field of data will really help our students. We appreciate you talking to us and informing us on some very interesting topics. Thank you.  

Justin Kyle Bush  19:35 

Thank you for listening to Epod. For more episodes, visit interpro.wisc.edu/podcasts. And if you enjoyed this, don’t forget to subscribe, rate, review and share.