Making the Business Case for Machine Learning

Host David Broussard welcomes Alberto De Obeso and Zach Garner in a rich conversation about why and how to implement machine learning at your organization.

GUESTS: Alberto De Obeso, Solutions Lead; Zach Garner, Lead Consultant 

HOST: David Broussard 

DESCRIPTION: Host David Broussard welcomes Alberto De Obeso and Zach Garner in a rich conversation about why and how to implement machine learning at your organization. 

Adelina Kainer   

Welcome to Digital Reimagined, a podcast packed with insights from Apex Systems, a world class technology services leader working to reimagine value for our clients. 

 

David Broussard   

We'll bring you the voices of industry experts to showcase our proven solutions that span across digital innovation, modern enterprise, and workforce mobilization.  

 

David Broussard   

My name is David Broussard, and I am one of your co-hosts for the Digital Reimagined podcast for Apex Systems. I'm joined in the studio today by Alberto De Obeso and Zach Garner of Apex Systems. Guys, I'm super excited to be here with you today. Can you give us just a quick intro on each of yourselves? 

 

Alberto De Obeso   

Sure. I'm Alberto De Obeso, I am a solution lead for AI. So I started as a developer and now I'm focused on data engineering and data science.  

 

Zach Garner   

Thanks, David. I'm Zach Garner, I'm a lead consultant at Apex Systems. I specialize in machine learning, computer vision, and data engineering. 

 

David Broussard   

Alright! Well, the topic today is building a business case for machine learning. I think as many of us are seeing out there in the consulting and sales world, it's what everyone is talking about right now. 

 

Zach Garner   

I think we're in an amazing time in machine learning in the industry. We're kind of getting past the initial hype phase. We're getting past the point where you have to have a team of PhDs to do machine learning. You could have a team of machine learning engineers, some data scientists come in, understand a problem, and build a solution in a faster timeframe. Just a few years ago, many of the machine learning tasks had fairly low accuracy, took a lot of work, took a lot of data, took a lot of research, and now you're able to generate value in a much faster timeframe. 

 

David Broussard   

That's awesome. Thank you for sharing. And, I'd love to hear your thoughts as leaders in the practice around what our clients saying right now? What are they talking about when it comes to these concepts? 

 

Alberto De Obeso   

Well, I'm actually really interested in optimization. It's a case where you take a process and then automate part of this process. So this is called robotic process automation. And it's really interesting, because now we can mix different AI disciplines into this. So if the machine is to perceive something complex, we can embed some image processing, then the machine can see. But then, some text is always involved. So we can make the whole process flow with some text understanding. We can understand what the text says, and then we can increase the complexity of processes. In the end, what the client gets is more fluid processes and an interaction between people and machines. 

 

David Broussard   

Awesome. Zach, what's something that's exciting you about your projects that you're working on right now? 

 

Zach Garner   

Sure. My clients lately have been mostly in oil and gas, industrial farming, places where machine learning is now making inroads outside of the digital-first organizations that were the first movers. There's a big push towards predictive maintenance or predictive ops. This is taking some concepts like anomaly detection even further. So what's happening is these big refineries or factories are installing IoT devices, sensor devices. They're reporting back all kinds of data about these very expensive equipment, very expensive large-scale factories. And we want to not just identify when a problem occurs, but to forecast out before that problem happens. And the idea here is that if you've got an oil refinery, you want to know before a problem happens. And when you're sending out a specialist mechanic to go fix it, you need to be right. So these problems require machine learning that will analyze hundreds, thousands of IoT sensors, and make sense of that. This is where deep learning succeeds.  

 

David Broussard   

Fantastic. So with companies having changed so much over the past 12 to 15 months, a lot of it forced upon them because of the pandemic, and some probably moved into these AI and machine learning phases before they were ready. What are the steps an organization needs to take to be ready to implement AI and machine learning? 

 

Alberto De Obeso   

Okay, the first step is to actually have some maturity in the data. The input of every model is quality data and oftentimes we see that they are not ready because we analyze the quality of the data sets and we find that we don't have enough completeness to actually do some modeling. For instance, if you want to get a view in the future of when your machines are going to fail, we will need some data in the past with some quality, and the model will need to be fed data frequently. So if we don't have the pipelines to provide that, it's really hard to get a model that will work in production. 

 

David Broussard   

Thank you for that, Alberto. Zach, any other thoughts on that concept? 

 

Zach Garner   

Alberto is right on. I think another thing to consider is there's also the culture of the organization and the business processes involved. You have to be able to collaborate effectively across a number of domains in the organization, both technology-wise and business, getting subject matter experts out of the field and connected with the machine learning engineers and data scientists. And ultimately, that takes a bit of a cultural shift for an organization.  

 

David Broussard   

That's great words to hear, Zach, just thinking about these transformations that organizations are going through and the change, the disruption that it can cause. And the ability to get the organization to buy into those changes is so impactful for any type of initiative like this. So Alberto, Zach, when it comes to the impacts that implementing AI and machine learning can have on an organization from a positive standpoint, what are some of those? And how can Apex Systems help companies achieve some of those? 

 

Alberto De Obeso   

I think one of the most interesting avenues is the implementation of robotic process automation. So once your process is automated, then it can run freely, and then fulfill some of the client's needs. I think that's one of the greatest opportunities. 

 

David Broussard   

Great, how does Apex Systems play into a customer's journey in RPA and improving user experience, Alberto? 

 

Alberto De Obeso   

I think we have two main options. The first one is the creation of a center of excellence. We call it a robotic operations center. So we understand your company and then we create a plan for increased governance, documentation, good practices, to actually know the full lifecycle of the robot, and then provide standards. And the second part is we provide you with a footpath. We have a financial lead, we also have experts in RPA engineering, also QAs and BAs. So we can extend these kind of teams to actually do the implementation, following the good practices that we establish during the creation of the robotic operations center. 

 

David Broussard   

Awesome. So whether you're at the beginning or somewhere close to the end, it seems like there's a place that Apex Systems has the ability to add value to your journey. So final thoughts: if I'm the leader of an organization considering the implementation of some of these processes and procedures, what's something to start those thoughts and figure out if I'm ready? 

 

Alberto De Obeso   

Okay, I think it's really important to understand what your staff does everyday, because AI enables the synergy between machines and people. So it may be the case that you, observing what your staff do, you can see that a portion can be delegated to a machine. So the reality is that as time passes, we are able to automate more complex tasks. So this means that this chunk of work that the machine can do is increasing. I would say, start by considering, "What will happen if I am able to automate 20% of the current work? What will be the impact?". Chances are that we at Apex will be able to help you to automate that 20% of the work. So that means a lot of opportunities. But I am sure that if you analyze your current work stream, you will find out that a percentage of this work stream can be automated. So I think that would be a good starting point. 

 

David Broussard   

That's a great thought, to think about what's out there that can be done without someone who can go spend time doing something more impactful. It's great. Zach, anything from you, sir? 

 

Zach Garner   

So machine learning is data science at high scalability. It's looking for more complex patterns, high volume insights. It's taking things to a next level. So look at your data analytics platform, your data engineering platform, talk to your data scientists, see where you can take things to that next level. Start to find out the things you don't know about your data, the limits of that knowledge, and see where it can be applied and leveled up. 

 

David Broussard   

Well, Alberto, Zach, it was great to chat with you about this concept. Appreciate both of you being on here and looking forward to doing more of these with you on some other topics in the future. 

 

Zach Garner   

Yeah, it was a blast. Thank you. 

 

Alberto De Obeso   

Thank you. Thanks a lot. 

 

David Broussard   

Please be sure to subscribe to Digital Reimagined, wherever you listen to podcasts. 

 

Adelina Kainer   

To learn more about Apex Systems' offerings, visit us at apexsystems.com/insights. You'll find our podcasts here along with success stories, articles, news, and trends. 

 

David Broussard   

The music you heard was Do Ba Do by Otis Galloway.