Best Practices in the Public Sector for AI/ML in 2020

Today we are visiting with Eric Clark, Technical Project Manager & Cloud enthusiast at SpringML. Let’s listen in as he shares best practices for the public sector and civil projects as well as a customer story around what happened when the City of Memphis asked what can I do with Machine Learning.  

Episode Overview

Eric Clark is a Technical Project Manager & Cloud enthusiast at SpringML. In this episode, Eric shares best practices for the public sector and civil projects as well as a customer story around what happened when the City of Memphis asked what can I do with Machine Learning.    

>> According to the International Data Corporation, the rate of data growth is expected to hit a hundred and seventy-five by 2025.

A decade ago, integrating disparate data sources to drive insights was an expensive and painful undertaking. Fast forward to 2020, there is a lot of opportunity for the public sector to leverage data analytics to solve problems quickly.

“The growth of data and what’s coming into organizations can be either a good thing or a really bad thing? It’s good if you know how to handle it and have a plan. But it can be very bad if you’re just not prepared,” Eric explains.

During the episode, Eric shared how SpringML helped the City of Memphis leverage the data they already had to detect potholes across the city and in doing so unlocked the art of the possible.

For more of our deep dive into best practices in the Public Sector for AI/ML in 2020 listen to the full podcast episode.  

Episode Transcript

Tamera – Welcome back to another episode in SpringML’s podcast Vision 2020. Today we are visiting with Eric Clark, Technical Project Manager & Cloud Enthusiast at SpringML. One of his many roles at SpringML is helping customers get to the Cloud and unlock the out of the possible in their AI/ML Adoption Journey. Let’s listen as Eric shares best practices for the public sector in civil projects as well as a customer story around what happened when the City of Memphis asked what I can do with machine learning? So, Here’s your host, Chief marketing officer Megan Dahlgren.

Megan –  I’m so excited to introduce Eric Clark, a technical program manager, and leader at SpringML who helps our customers get to the Cloud and increase their comfort level with experimentation. Today, we are talking about AI/ML in the public sector, state and local government which is traditionally known to be a late adopter in Innovation adoption. Is this fair to say?

Eric – I have been reflecting on what I’ve seen over the past 10 years and thinking about what I was doing? I was working for a company doing consulting for federal agencies. We were building a project to integrate almost 2000 data sources. One of the largest data warehouses I’ve ever seen. I was just thinking about what it took to pull that off back then and compared to what I think it would take to pull something like that off today. Those two views are so different. It was then very hard and very painful to get useful insights out of data. Painful to bring it all together and a costly process. Lots of people involved, a lot of change requests and people just very narrowly focused on data integration and building reports requirements on what had already happened.

It wasn’t predictive at all. So all of that effort was a painful process. Today, there’s so much more opportunity for the public sector to take advantage of data analytics and a much easier way to solve much more significant problems quicker than ever before. We are pretty excited about what we see in this space.

Megan – Reading between the lines, I hear you say that you needed a strong business case to get more out of your data. With lots of inputs, constrained budgets, and longer project cycles, Cloud opens up the art of the possible.

Eric –  I did a presentation four years ago in 2016, and happened to pull some statistics from the international data corporation. Every couple of years, they publish numbers about the global rate of data growth. We were at about 15 zettabytes in 2016, and they expected about 40 zettabytes by 2020. In 2018, we were already at 33 zettabytes and expected to hit 175 by 2025.

The growth of data and what is coming into organizations can be either be a good thing or a bad thing. It’s good if you have a plan to handle it. But it can be very challenging if you’re not prepared or don’t have the staff, skills, or the tools ready to be able to handle large amounts of data.

Megan –  I understand that being a Cloud enthusiast and doing projects for customers excites you. Could you help walk us through how you’re doing it for your public sector customers?

Eric – On the project with the City of Memphis, the engagement started with a pretty important question. What can I do with artificial intelligence and Machine Learning? We wanted to start with a proof-of-concept to answer that question. The city of Memphis had an incredible amount of data and videos. 

We came up with a use case to fix one of the biggest problems they were facing, detecting potholes across the city. It’s an essential thing for the city to understand where they are and how to repair them? There wasn’t anything that we had to do in terms of installing equipment. We accessed their data sources on the city bus system. All of the buses had front-facing cameras and we used the video and trained machine learning models to watch the video and pick out potholes.  We used GPS data to tag every time it found one. This helped in addressing issues quicker and contributed to optimizing routes for the people in the cruiser filling those potholes. It’s an excellent use case. 

Megan – Can you give any best practices about how the public sector can get started about thinking through this to kick off their analytics journey?

Eric – There are a lot of best practices like data governance, breaking down silos, getting access to different data sources, and having an initial AI/ML strategy. You should think about defining these first. Focusing on the infrastructure that you need to run machine learning models or advanced analytics is a top priority. 

Adapting the Cloud naturally gives way to focusing more on the results. It is also better not to bite off too much at once.I know it’s easy for people to dream up a hundred different use cases. However, you have to start somewhere small with a clear use case. Don’t try to do all of them at once. 

Lastly, I would mention the talent shortage here in this space. It’s tough to find a sound data engineer’s or good data scientists because there aren’t a lot of them. So I would recommend especially in the public sector, if you see potential in your existing staff,  figure out what you need to do to enable them. 

Megan – Do you have any examples on getting how best to get started?

Eric –  Yeah, I think it’s straightforward to get started with an initial simple machine learning model that does a simple detection from your images or video or a predictive model. There’s a lot of publicly-available machine learning models that you can consume. For example, you could go to Google Cloud right now, spin up some infrastructure within minutes, and use some of their pre-trained models to do detections or interesting predictions on your data. If you can set the goal to enable curious people by giving them some access to the data, you’ll see some exciting things happen. 

Megan – Are there some use cases or problems that you frequently see in the public sector?

Eric – There’s always data silos. When folks are adopting Cloud, it becomes easier to break down those silos and start sharing some of that data cross-agency. Another project example that we are working with a state agency, which will have a significant impact on the citizens of the state, is synchronizing Social Services caseloads. What we’re trying to do is consolidate a bunch of different data sources from various Social Services programs across the state like workforce placement, department of corrections, reentry programs, etc. The end result will reduce overlaps, reduce the stress for the citizens receiving services from multiple programs. 

Megan – Well, that sounds like an ambitious program. At the same time, it seems like the Cloud makes it much easier to do ambitious projects.

Eric – 100% True. It’s all about teaching computers how to understand data. There’s no way that humans are going to be able to process the vast amounts of information that is increasing every year. We have to teach computers. That’s where AI & ML starts to unlock that potential.

Megan – I can’t wait for our next interview to talk more about it, look into some more of these use cases. Thanks so much for coming Eric.