All right, welcome everyone to the Innovation Flow podcast where we talk about all things park related innovation, pilot studies, research that's going on at South Platte Renew. And we are on this floor, the vendor floor here in Chicago, weftech. And we are here with Jamie Lefkowitz. Thanks for being here, Jamie.

Thank you.

Jamie is the innovation development lead at Brown and Caldwell and she's here to talk to us a little bit about. Well, I'll let you tell about it. It's about PFAS and source tracking of pfas. Before we get into that, why don't you give viewers a little bit of a background on who you are, where you came from, how you got in the business. Jamie?

Sure, yeah. I work at Brown and Caldwell. As Blair said, I've been in the business for almost 20 years now, mostly doing consulting, but had a stint at a startup, a tech startup, which really changed the trajectory of what I wanted to do. Really got me into innovation, digital technology. And over the past few years at Brown and Caldwell, I've been building a data science team. That's where this connects to this project. Using data science, really, to make better use of water data. There's a lot of data out there and utilities and others aren't necessarily using it as well as we could to draw conclusions, to do research. So that's my aim right now, is to make better use of our data.

Nice. Yeah. That is a hot topic in the wastewater industry these days. We've been talking about recovering resources. People think of gas, people think of biosolids, nutrients, things like that. But another area is data and information that you can gather both in the wastewater. They're even doing like COVID testing in wastewater.

Yeah, we did some of that. Huge. It's just so many opportunities to look at the data that's in the plant, in the collection system, in the customer base, really to manage the water better and to inform other decisions and policies.

Yeah. Cool. Well, how's your conference here in Chicago? Have you been doing a lot of things here?

It's been great. Yeah. I love the city. Come back here every couple years, really, for this conference. I'm from California, not the big city part of California, so it's nice to be able to walk around and see the big buildings.

Yeah. Did you, you said you went to the Bean. I've been asking everybody, did you go to the. What do you think of the Bean?

It's super cute.

It is, isn't it? It's like built for selfies. It's like they made a Thing just to take selfies.

It's great. It's really hard not to be drawn to it. You see it and you're like, I want to go stand near that. It's very cute.

I love it. Yeah. And then you're like, what is this? But then you see yourself in the thing. Like, this is awesome.

Yeah.

Yeah. Cool. Well, yeah, thanks for being here this morning. And I want to get a little bit more into this project that you have been working on with South Platte Renew, as well as of Oklahoma, I think, is the partner.

That's it.

Why don't you tell us a little bit about what the project is, what the aim of it, what the goals are.

Yeah. So our goals with the project ultimately are to make it easier for utilities to track the source of PFAS contamination that's in their influence. So wastewater utilities do not generate PFAs. They accept the contamination in the waste stream.

PFAS being. Describe PFAS for listeners who are. I think everyone's familiar by now because it's a hot topic.

If you haven't heard of pfas, they're the forever chemicals. So used in many consumer products and manufacturing materials, and it makes things slippery. Think Teflon, think Gore Tex, think waterproof, miracle chemical turned forever chemical. Most of them do not break down. And so they persist, and they persist in our water cycle and just keep making the rounds. So we're learning that they are also very bad for people.

So this is how we do it. Things like, what am I thinking of? Ddt? Things like asbestos. You know, when they come out, we're like, this is the greatest chemical. And then 50 years later, people start to say, hey, we didn't think this through, you know. Yeah, so, yeah, let's get back to the project. So how does this project combat or try to track down that forever?

Yeah. So like I said, wastewater plants are not generating PFAs. They are accepting it in the waste stream. So in order to control it and ultimately eventually reduce it, they're charged with permitting and finding out where it's coming from. And PFAS comes from a variety of places pretty much everywhere at this point. But if you break down the wastewater stream, you have domestic wastewater, and yes, there's PFAS there. Then you have your permanent dischargers. And some of them do discharge PFAS into the influent, into the collection system. Then you have some distributed sources, commercial sources. And so one of the things that utilities need to do is track the source so they know PFAS is coming into their plant. But from where is it coming? And so what this project is doing is building on previous research where we've been able to develop machine learning algorithms that help us track the source of PFAS in the environment.

Nice.

So we take samples from groundwater, rivers, lakes. We're able to say that source came from a landfill. So the pfas in that water, this is landfill pfas or this is old firefighting foam pfas.

So just like a fingerprint on landfill would have a fingerprint PFAS signature.

Yes. And that signature is in the data and it shows up in the component concentration. So PFAs, maybe the listeners may or may not know it's not one compound, it's hundreds of compounds. Making it even more complicated. But what that means is that you can measure the levels of those compounds. And our hypothesis is that that's a fingerprint. And you can, if you have enough data to train a machine learning model, that model can then what's called deconvolute, but it can find that fingerprint.

Deconvolute. I like that. So is the idea like tiny things humans might not be able to see looking at a lab report, AI or machine learning can tease out?

Yeah, and you can imagine it takes a lot of data.

So.

And it's not that humans can't see it. We can't look at 13,000 samples all at once and find the patterns in them. And we look at it and we see, oh, well, this, you know, this metal plater doesn't really look like this metal plater. And then we get an unknown sample and we're like, well, I don't know, it could be anything. But machine learning algorithms are really good at finding patterns in data. That seems like it's hard to do, but when the machine learning algorithm gets a hold of it, you can do a lot of computations at once and find those patterns and then feed them back to us. Cool.

Well, I know there's the fingerprinting library component and then there's. I think what you described, I've heard before, is like an early, early warning system attempt to build that. Talk about that side.

Yeah, so it seems a little disconnected, but at the same time, another thing happening at the influent is the PFAS levels are going up and down. So it would be nice to know when you've got high levels of PFAS coming into your plant, you can then do some sampling. You can then start to do that source tracking. So what we're doing is Testing. We're in the very early stages of testing to see if we can find a sensor that can indicate when PFAS levels are high or low. We don't think right now there isn't a sensor for pfas, unfortunately. But we think that maybe some of these technologies we might be able to find CO contaminants or like I said, an indication of binary. High PFAS or low pfas. Very early stages.

Yeah. What challenges? This is the blending of technology data. But then there's raw wastewater. Is the subject material even before it gets to the wastewater plant? Some of this coming out of industries. What challenges has that posed or has it to this project?

I mean, the number one challenge is we don't have enough data in the right place. And I started this with saying we have so much data, we just need to use it. So we're working with SBR on this. You guys have done a big source study and collected a lot of data points, but in the big scheme of things, that's not enough data points to build a training data set for a robust model. So we're trying it out, but what we need is data from all the utilities. And there are a lot of utilities collecting PFAS data from up in the collection system.

Yeah.

So we need more labeled data to train the models. And then like you said, wastewater is hard. It's hard to deal with. It's very dirty. It's hard to put sensors in it. It's actually to sense the PFAS at low enough levels all the time with the methods that we have. So just collecting that data has been the most challenging.

Yeah, I could see that. I can totally see that. But if it works, what are the outcomes? If it all comes together, you can have this signature and you can have a probe that says, hey, PFAS coming in. What does that look like?

That's the vision. Right. The dream is to be able to have a light that goes on when you have higher PFAS coming in than you expect, can go out. You can take a sample. Take that sample. It's your fingerprint. You can put it through a computer algorithm and what that will spit out as your most likely sources.

Yeah.

And then you can pick up the phone and call your customers.

Fred's plating company, Right.

Hey, did you guys do something? And that will help you eventually deal with the issue, you know, whether it's bypass or treatment, and we're not there yet, we're probably never going to be able to remove all the pfas. But source control is going to be a big part of it. Yeah.

Yeah, I think so, too. Even in residential sources. The more we can educate, the more we can. And South Platte Renew has been working with the legislature to pass a ban on pfas, you know, chemicals or products that PFAS added. If we can get this stuff, that's the key. Get it out of the consumer stream so that it's gone. It may take 20, 30 years, but I think ultimately that's what's gonna have to happen. But in the meantime, there may be treat or product substitutions and things you can do to get rid of it.

Right.

Well, cool. What's been the most fun part about this project for you?

The most fun part? Let's see. I think learning about it all PFAS is not my expertise, not my background. Like I said, I built a data science team at Brown and Caldwell, but learning as much as I have to about PFAs, fate and transport, and working with our professor partner at the University of Oklahoma, it's just really cool to see all of this coming together and particularly partnering with our utility on this. So we're not just doing this in academia. We've got the real utility in the room all the time making this happen. Also want to thank the Water Research foundation that's funding this as a tailored collaboration.

Yeah, we should mention them, which we should.

And the tailored collaborations are just so cool because it's always the start of something bigger when the utility and WORF come together and realize an idea that you all have. We love being a part of that.

Yeah, yeah. And I like the kind of the idea that this is a start, this is beginning. We're learning we need a ton more data. We're learning we need more robust probes or technology on that end. But you don't learn that if you never start something.

Exactly.

Cool.

Well, tell me a little bit more about this data team that you've set up at Brown and Caldwell. How has that been? How is integrating a data team into a bunch of engineers and scientists? How's that been?

It's been great. First of all, the team we set up, they started out as environmental engineers. And so we've got some young staff who are passionate about data science, about machine learning, about AI. And so Brown and Caldwell supported them learning even more, really becoming data scientists. It's quite rare that you even have people coming out of school with a data science degree at this point. So we took the path of turning our environmental engineers.

Grew your own.

Yeah. So we grew Our own. And they are just. Each of them has picked kind of the thing that they're most passionate about and pursuing. So we've got people working on large language model uses in the industry. We've got people working on time series predictive analytics in wastewater treatment and in water resources planning. So it's been really cool and the response is great. We get to work on really cool projects like this, and it's just a very dynamic environment in our research and innovation team.

Yeah, it's crazy how much things have changed. When I got out of school, it was like Excel spreadsheets. That was high technology, trying to get away from those Python power bi. Just like the kids, I guess, kids, people coming out of school now, they're like, they just know tech. Tech's part of the curriculum.

It's native to them. Right. It's almost like everyone gets a little software engineering in college now, and we want to use that in this industry.

My son is a freshman CSU and his the hardest class, but I think he likes his coding. And I'm like, they're making you take a coding class?

That's great. Right?

So. Well, that's great. Thanks for. Thanks for coming by and sharing some of this info. I think it is a great project. Get PFAS out of the stream by identifying where it's coming from and when it's coming in. And the idea of using data and machine learning and people focused on this to get it done is great. So thanks for coming by and sharing that.

Thanks so much for having me, Blair.

You bet. Welcome everyone to the Park Innovation Flow podcast location here at WEFTEC, the exhibition floor in Chicago 2025. And I am here with an innovation leader. Mahesh Lunani with AquaSight is my guest now. Thanks for being here, Mahesh.

Thank you, Blair. Pleasure to be here.

Yeah.

Mahesh, I wanted to talk to you. Thanks for coming by. I wanted to talk to you. You know, PARC is all about innovation. PARC is about research data. How do you fit into the innovation space? I know you got a lot going on in that area.

Well, first of all, I want to congratulate Peter, yourself and the whole team. What you guys doing at PARC is amazing.

Well, thanks.

And I think once you hit certain milestones, truly the rest of the country can replicate. So I congratulate you for that.

Well, I appreciate that.

Yeah, yeah, no, definitely. And then as far as aquisette is concerned, concern, as we know, data is power and utilities have no dose of generating data. They generate enormous amount of Operational maintenance, environmental data. What we see with the advent of AI is to be able to make that very, very efficient, to be analyzed, use it to predict and forecast. But that. That's what we call algorithmic AI. The next jump is the generative AI where you can actually utilize large language models to be able to auto interpret these data sets and to be able to provide insights, what's happening, reasoning as to why it's happening and actions as to what to do with it. And that I think is a holy grail.

Yeah.

And this at vtec, we launched a water assistant exclusively for the water sector. Wow. Called Ava.

Ava.

We are excited about it.

Yeah. How did you pick the name Ava? Is it a family name or.

No, I, you know, there's all. First of all, it's very easy to remember.

Yeah.

Right.

But more important, it's on a button.

It's on a button. That's right, Ava. But more importantly, it's. It just seems like it's something an advice you can take from. It's very simple and easy to remember. But this has been in plans for three years in the company. But the advent of large language models accelerated our ability to offer this to the entire industry.

So if I'm a wastewater plant, wastewater operator, what kind of things can Ava help me me with? So Ava, be my assistant.

Excellent. So if I'm an operator, as you said, if you're an operator, you care for a few things. You care for how your process is working.

Oh yeah.

Like svi, sludge volume index. And is it within the best practice? Right. So one thing Ava can do is it auto feeds the data around sludge volume index and all of the parameters are impacted by sludge volume index. It'll tell you how to operationally maintain a steady optimal state. Typically, you want the SVI to be between 80 and 100 or lower so the sludge can settle. What it does is it. First of all, it's doing the thinking and providing. So it's not stressing out the operator in terms of remembering everything. But more importantly, it's upskilling the operator to perform at the top quartile.

Yeah.

And that I believe is the power of AI upskilling the workforce.

I love it. I mean, I've seen throughout my career, just like you said, these plants collect so much into their SCADA systems through their probes, sensors. But it's like, what are we doing with it? And a lot of times it's nothing. So I like, you know, actually doing something. But then this, like you say, is next Level.

Level.

It's taken chat GPT, you know, that kind of concept, but for. Specifically for wastewater. Huh.

It's very interesting. You said this ChatGPT is too generic.

Yeah.

To make ChatGPT work or any model work, these foundational models work, you have to be content specific, context specific, and sector specific. And that's what we did with ava.

Yeah.

You know, as you know, most operators or utility engineers are very proud of what they actually do because it's complex. And they say, well, that doesn't apply to us. We've taken that out. It does apply to you because some other peer has done it and it highlights how other peers and the best practices have worked.

Yeah.

So now it eases the ability for a particular team member to accept and adopt because it's been done.

Yeah. That's interesting. This is. This is groundbreaking. I haven't heard of anything like this before, so. Yeah, thanks for. Thanks for telling us about it today. I also wanted to talk to you. I know, thanks for being on our podcast, but I know you also have a podcast where you interview people in the wastewater industry leaders. Utility director. Can you tell us a bit about your podcast?

Yeah. So the podcast is called 21st Century Water. There are about 44 episodes. It's been running for last four years, once a month. All utility leaders are the guests on the podcast. It's about. First of all, it is amazing for me to sit and listen these utility leaders, why they chose water, how they manage their business, how they handle their boards and communities, how they run a complex infrastructure under the constraints of regulatory pressures and cost pressures, and where they see the innovation and what they want their legacy to be. So for me to hear these stories and to share these stories to the entire water wastewater communities, and we have actually very good, strong following base on those that listen to this podcast. Peter has been on this. It's gonna release in about exactly less than two weeks. Peter's episode, I think it's episode 45. It's on eight different platforms. It's free, so from Apple to Google to Spotify to iHeart, etc. And it's for me, one of the. The most gratifying moments. Just listening and all I do is ask questions. Yeah, not even that complicated for me.

I am, I'm a listener myself. And I like what you're doing. I like the. You get a good, high caliber, quality guest, you know, as far as industry leaders from across the nation. And I think, I think it's great. I enjoy it.

Thank you.

Thanks for doing that. And thanks for being on our podcast here on innovation. Flo, did you have anything you wanted to leave the. Leave the viewers with before we close out here?

Yeah. No. So first of all, I believe we are at a very critical point in our industry where innovation, I'm not talking treatment technologies, but I'm talking digital and AI stuff that I deal with every day. And I speak to 200 utilities a year. Like 200 utilities. So really grand level. We are at a point where the performance of the water, wastewater, utilities can be dramatically upskilled. The workforce can be upskilled if we just embrace and deploy these as part of our daily business.

Yeah.

And I think I am excited what the future holds for the sector and never been more excited about it. So I'm glad to be part of this podcast.

Cool. Well, I'm excited with you and thanks again for coming on, man. Mahesh, I appreciate you being here to our viewers. Thanks for watching Park Innovation Flow Podcast. If you like the episode, give us a five star rating on whatever podcast platform you're listening to or YouTube channel and we'll see you next time on the Innovation Flow Podcast.