When legacy systems failed during critical union votes, APFA’s Todd Smitala led a high-stakes real-time data transformation. Learn how Confluent Cloud enables a data streaming platform to support 27,000 flight attendants with speed, scale, and reliability.
What does real transformation look like when the stakes are high? Todd Smitala, Technology Solutions Architect at the Association of Professional Flight Attendants (APFA), shares how his journey of modernizing data system support 27,000 union members.
With the legacy infrastructure crumbling under peak traffic during a high-volume union vote, Todd’s team turned to cloud-based real-time streaming with Apache Kafka® on Confluent Cloud. The result? A more reliable, scalable system that delivered immediate impact, ensuring every vote counted. In this episode, Todd walks through the technical decisions and leadership alignment that made it possible.
You’ll learn:
If you're leading a data streaming initiative and need a blueprint for impact, this episode is for you.
About the Guest:
As a Technology Solutions Architect at APFA, Todd Smitala brings over 10 years of experience as a flight attendant and 20 years in IT, allowing Todd to bridge the gap between technology and the needs of flight attendants. He’s passionate about leveraging technology to create meaningful change, support union members, and enforce contractual legalities. This combination of industry insight and technical expertise empowers Todd to deliver innovations that make a real difference. Outside of work, Todd enjoys scuba diving, adventure travel, hiking, playing music, and dancing at Carnaval in Brazil.
Guest Highlight:
“We all agreed that Confluent Cloud would be the way we want to go because of how it does host. It does really handle all of the background stuff. I'm not a Java developer… We didn't have the time to invest in all of that so Confluent Cloud is actually perfect for this type of thing.”
Episode Timestamps:
*(01:22) - How APFA Is Using Data Streaming to Empower Members
*(20:39) - The Runbook: Tools & Tactics
*(26:45) - Data Streaming Street Cred: Improve Data Streaming Adoption
*(28:30) - Quick Bytes
*(33:36) - Joseph’s Top 3 Takeaways
Dive Deeper into Data Streaming:
Links & Resources:
Our Sponsor:
Your data shouldn’t be a problem to manage. It should be your superpower. The Confluent data streaming platform transforms organizations with trustworthy, real-time data that seamlessly spans your entire environment and powers innovation across every use case. Create smarter, deploy faster, and maximize efficiency with a true data streaming platform from the pioneers in data streaming. Learn more at confluent.io.
0:00:00.2 Joseph Morais: Welcome to Life Is But A Stream, the web show for tech leaders who need real time insights. I'm Joseph Morais, technical champion and data streaming evangelist here at Confluent. My goal, helping leaders like you harness data streaming to drive instant analytics, enhance customer experiences and lead innovation. Today I'm talking to Todd Smitala, Technology Solutions Architect at the Association of Professional Flight Attendants. Representing over 27,000 flight attendants, APFA plays an important role in making sure its members' voices are heard loud and clear. But what happens when outdated tech gets in the way? That's where Todd comes in. Our guest today shares how building a cutting edge data streaming architecture achieved incredible results. You'll hear the tools and strategies Todd used to make it happen. But first, a quick word from our sponsor.
0:00:55.3 Announcer: Your data shouldn't be a problem to manage. It should be your superpower. The Confluent data streaming platform transforms organizations with trustworthy real time data that seamlessly spans your entire environment and powers innovation across every use case. Create smarter, deploy faster and maximize efficiency with the true data streaming platform from the pioneers in data streaming.
0:01:22.4 Joseph Morais: Joining me now is Todd, Technology Solutions Architect at the Association of Professional Flight Attendants. How are you today, Todd?
0:01:28.5 Todd Smitala: I'm doing great, thanks Joe.
0:01:30.2 Joseph Morais: Excellent. Well, let's jump right into it. Tell me about the APFA.
0:01:34.7 Todd Smitala: Okay. We are the union that represents the flight attendants at American airlines. There's about 27,000 of us right now.
0:01:42.2 Joseph Morais: Fantastic. So tell me particularly what you and your team do at APFA.
0:01:46.9 Todd Smitala: At this point I do pretty much anything technology related. I was brought in just after COVID with a new leadership team that came in. It was some young blood that was coming in as well and they wanted to do a little bit more focus on data and tech revamps, things like that. So I was brought in, I'm a flight attendant as well. So I was brought in to bring a flight attendant perspective to a website redesign. And at that point we were also heading into negotiations and they wanted to kind of take a different approach to negotiations this time. Usually that whole process is completely secret and silent and we don't know anything until they actually come up with an agreement and then they present it to us and we have 30 days to read 200 pages and decide if we want to vote on it or not. So this time they said no, we want to put on the website, we want to bring this information out, we want to work with social media, we want to make sure everybody sees every step of the way until we get to that.
0:02:40.7 Todd Smitala: And so I was in charge of making sure that happened. And that's how I got started there. And that really brings us into the data streaming thing. The big project that I've been working on these past few years is really a member database that when I got there was on an enterprise server in the closet and has all the information. You know, that's all...
0:03:00.6 Joseph Morais: Have been there before.
0:03:02.7 Todd Smitala: You've been there, right? It's from 2014. It was, it's fine. You know, we had up until that point it was just people that were in the office, a few people interfacing with that. It was okay. But at this point what they wanted to do, not only revamp the website, but they wanted to use that enterprise system. It has a web portal as well. They wanted to bring that out and make that available to all 27,000 flight attendants. So I immediately knew that we'd need to take that one out of the closet and put that in the cloud. That was...
0:03:32.1 Joseph Morais: Right before it would be taken out of the closet and put it in a server rack, like in a proper server room. But now you just go right to the cloud. You got to skip that.
0:03:37.2 Todd Smitala: You want right to the cloud on this one because of the longer term things we wanted to do with this. Right. So they wanted to have a single sign on experience. They wanted people to be able to go into our WordPress website and then also be able to access that member portal and not really know that they're in another phase and another platform, anything like that. So.
0:03:58.0 Joseph Morais: They just want it to look seamless.
0:04:00.9 Todd Smitala: All seamless. Exactly. So I was coming from, I'm a developer, so I was coming from a React and Node and Ruby on Rails type of thing. So I was thinking immediately of OAuth, you know, that would be a nice way to handle the identity management and all of that. So we sat down with Okta to see how this might work and we had. Our WordPress site was not a regular WordPress thing. You know, it was really, it wasn't a simple thing that you kind of, you know, any person can go in and create their website. It had to become so specialized because of the limits to certain areas for our members. We had to have a team actually handling it. So there is a WordPress team that does that part and then we have the team that's handling the server side as well, any implementations there. So all of us got together and everyone thought well Okta's perfect, let's do it. But then the cost came in and you know how that goes. So the leadership said, we can't do this. We can't afford this plan B, which was, okay, let's work with these two teams and we'll come up with something that I call Frankenstein.
0:05:01.3 Todd Smitala: So we have this little, this login means of logging in and having that single sign on experience between those two very specific two platforms. And it worked okay initially until we had our first big boat, the strike vote. So we had a number of big votes coming along. You know, our first one was a strike vote in July of 23. And we did expect a lot of people coming out for this vote. You know, we did a big campaign, we did town halls, we did a lot of Instagram, you know, a lot of emails, all that kind of stuff telling everybody, you know, you need to log in at this time. This is when the vote's starting. It's actually set to be a month long. But here is our mistake. We started the vote on the 20th, I believe the 27th or... It was 28th of July and then we ended it on the 29th of August.
0:05:54.8 Joseph Morais: Okay.
0:05:55.4 Todd Smitala: But a lot of people read just the 28th to the 29th. They thought it was a one day thing.
0:06:00.7 Joseph Morais: They thought they had one day.
0:06:02.4 Todd Smitala: So literally when it was time to log in for that, we had a good amount of those 27,000 people logging in immediately at the vote open. They thought it was only one day. That alone really overwhelmed that WordPress type of hosting server infrastructure that we had anyway. And worse than that, most of those people did not remember their login credentials. So they were all clicking forgot password. And that enterprise system, it's not designed for that at all. It basically collects any emails that need to be to go out for a two minute period and then the batch sends them. So you can imagine how that just spiraled. It was just an absolute nightmare. We had to do it.
0:06:39.9 Joseph Morais: This is a great setup. APFA's data streaming setup has accomplished some really impressive feats. I know we kind of set up how things were, but I know things got much better. So tell us, how did it affect your ability to engage with your members?
0:06:54.0 Todd Smitala: If we look at the results of that strike vote. It was, we had about a 93% turnout out of all of the ones that could go that could vote, and it was overwhelmingly yes. It was over 99% that said yes. So, you know, that was a pretty good turnout. That was something that we really wanted for that. And even with that Frankenstein setup, we still managed to get that. So that was important.
0:07:20.7 Joseph Morais: Okay, got you this. But then of course with that, all that back pressure from that day one that exposed the problems.
0:07:26.9 Todd Smitala: The problem is for when we have other voters, like for the actual contract, it's not a given. We really don't know. You know, people need to really understand that and make the proper decision. And so that's the one that we needed to make sure not only did we have a high voter turnout, but that it was easy for them to do. So it was extremely important for us to do that. So that's what happened after the last one. Our leadership team sat down with me and the President in particular said, you have full range, you know, you can do whatever you need to do here. Don't worry about the budget, just make sure that this works for the next time. And we knew it would be coming within six months to a year. So they said, you know, let just go ahead with that first plan if that's going to be the best one, whatever you can do to make sure that people are able to log in very easily, that there's not going to have this whole backlog of trying to reset their passwords, all of that kind of stuff, have an ease of access to be able to vote. That's what we want.
0:08:22.1 Joseph Morais: Walk us through what you built, you know, with all those requirements in mind to tackle this.
0:08:26.8 Todd Smitala: And you know, just taking one step back here. If you've noticed, we haven't said anything about data streaming at this point and it really hadn't, it hadn't even entered my mind at this point either. So that kind of gives you a background of where we're coming to and where we're going. So yeah. So yeah, then we're here. We go in with the setup. We've sat down again with Okta. We've, you know, with actually their Auth0 version of it now then we sat with those other teams as well. Everybody was on board it just, you know, we decided to go forward with it and in concept it all worked. But we were still thinking along these lines of how do we maintain information across these web platforms? You know, we're thinking, well okay, if somebody goes into the member portal, if they pay their dues, okay, this is all an API system as well. You know, it's not really going to work with a lot of 27,000 people doing things at once. But the problem became the enterprise system, you know, that's the brain of all of this, that's running all of these processes, that's the background of it.
0:09:32.4 Todd Smitala: So if that one, if there's an update there, it's not really directly connected to these web platforms, if there's something happening on those. So we sat with the Auth0 team again and they said, have you thought about Kafka?
0:09:45.7 Joseph Morais: Oh, here comes the data streaming client.
0:09:48.9 Todd Smitala: Where the first seed comes in. So have you thought about Kafka? We have. You know, we have some customers that have had similar situations and that seemed to work for them. You might want to explore that a little bit. And I had heard of it at that point, but I wasn't very familiar with it at all. And you know, we started talking with the other teams as well. Everybody seemed to think conceptually it was okay, but I was the one who knew the most out of it. So it was not, we weren't in a really good position at this point. But I did, I really dove into Confluent, in particular that, you know, there's a lot of information on there that I went through a lot of the tutorials and the videos and read and got through there. And we all.
0:10:25.4 Joseph Morais: I'm glad you found all of that helpful.
0:10:27.3 Todd Smitala: Yeah, we all agreed that Confluent Cloud would really be the way that we would want to go because of how it does host, you know, it does really kind of handle all of the background stuff because, you know, look again, I'm the one that's going to, that knows the most about this and I clearly don't have the background. I'm not a Java developer, you know, I don't have all of that going on in my background. So we didn't have the time to invest in all of that. So Confluent Cloud was actually perfect for this type of thing.
0:10:55.1 Joseph Morais: You just reminded me of something. Someone once told me this and I think it resonates, I think this will resonate with you. Someone said the only thing worse than writing code that you can buy is running a system that someone else will host for you, especially if they're experts in that particular thing. So I'm glad that what we build is exactly what you were looking for. So you don't have to think about how you run Kafka at scale or how you run stream processing at scale. You just had to go and do what you needed to do.
0:11:19.3 Todd Smitala: Exactly.
0:11:20.6 Joseph Morais: So you talked about Frankenstein, you talked about the new system that was built on top of data streaming. So now for the Big moment. What were the results of the new architecture?
0:11:28.7 Todd Smitala: Well, we did, we decided to go with using the enterprise system and we basically set up a specific table in that system that would track the different data points that we needed to maintain across all of the different platforms. So this is information like are they eligible to vote, do they owe any dues, what is their status, are they active, are they on leave, are they retired? These types of things.
0:11:51.1 Joseph Morais: You have to evaluate all these things right before you can give them a vote or not.
0:11:55.7 Todd Smitala: Exactly. All of that kind of together lets you know, are you able, are you able to vote? In particular, it's really, you know, are you current with your dues and are you eligible? We have some...
0:12:06.1 Joseph Morais: And are you doing that look up as they log in? Is that when the check is done?
0:12:09.8 Todd Smitala: Well, the information, the situation with them is happening all the time in the enterprise system. You know, there's all kinds of different processes that are happening nightly. You know, there's a billing cycle that happens once a month and if somebody goes in and pays, it all automatically happens there. So when they come in and login. Yeah, that's basically what happened.
0:12:29.4 Joseph Morais: It's all event driven, as it should be.
0:12:31.1 Todd Smitala: Yeah. So that's what we did. We set that particular table. We didn't want the entire database to be, you know, sending it out, we just wanted that particular table. So we set it up inside of the enterprise system that that particular table would have changed data capture heading, the events going on out to the Confluent source connector. So we got that information.
0:12:52.2 Joseph Morais: Right, so you had a connector, a CC connector pulling from that database, emanating those events into topics. Got it.
0:12:57.7 Todd Smitala: So exactly. Any change in one particular member's situation, if they've paid off their dues, it's going to send that event out. And so then we take that and we've connected that to Auth0. Now unfortunately, that was not direct. We didn't have a sync connection there. We had to, you know, bring a little mini Frankenstein in there of Azure functions and all of that.
0:13:20.5 Joseph Morais: But you got to keep your Frankenstein as small as possible.
0:13:22.9 Todd Smitala: Yeah. I mean there were things that we had to do. Like if there's a new member coming in, we need to create an account in Auth0. There were some extra things that we had to finesse around the edges to make it really work. It couldn't just be a direct connection. So that was all right in the end. But yeah, that's the infrastructure. It's going from that enterprise system into Confluent cloud and then from Confluent over to Auth0 and then Auth0 as it was designed, it kind of controls everything across the different web platforms and has all of that information right up to date. So if somebody wants to go in directly to the voting site to vote on something, it's going to have them log in, it will grab that metadata right away and it will know and it will be able to tell them, okay, you're not eligible and you can do this, you can click here and you can get eligible, you can make your dues payment or whatever needs to happen, they can go to their member portal, they can go to the website, whatever needs to happen there. And it worked out beautifully.
0:14:19.1 Todd Smitala: If you remember, I had said that on the strike authorization vote, we knew that a lot of people were going to vote for this and we knew a lot of them were going to vote for yes. But really, even in the end with that one, we only had 93% of the total electorate that voted. You can see right here on the 2024, this is our tentative agreement. When our big contract voted, we had a 94.43% turnout.
0:14:43.3 Joseph Morais: That's fantastic.
0:14:44.4 Todd Smitala: Okay. And out of that we had, you know, 88.64 that voted yes. So it was overwhelmingly voted yes. Okay. And then if we compare to the last one on the left hand side here, this is a different story, right? This is 2014. This was another, our previous tentative agreement vote. And this is the most telling part of that story. We have the results here. It was 8180 that voted yes and 8196 that voted no. A difference of 16 and they've the answer was no on that one. So that was a different kind of negotiation. It was a merger negotiation. So we didn't have that threat of a strike that we did now in this other one. So it's a little bit different. But really what happened in the end was it went into arbitration and they gave us the exact same contract, but they took away about three or four different prime features that we're still trying to get back now. So all because of, you know, if we had had a little bit better voter turnout, a little bit more, we wouldn't be crying about that now.
0:15:54.2 Joseph Morais: Right. It shows you how important that, you know, when it comes down to 16 out of 20, potentially 27 like that, you know, having every person's voice count is so important. And the fact that you were able to build something with data streaming that got you to over 94% participation is really incredible, Todd.
0:16:09.6 Todd Smitala: Yeah, we're very happy with how that all worked out. Without a doubt.
0:16:14.0 Joseph Morais: Very impressive result. Yeah. Like what were you hearing from your leadership? If you could not improve that member experience? If data streaming wasn't able to, you know, get you closer to 100%.
0:16:24.3 Todd Smitala: We'd run the risk of not approving a contract. We had already been. So think about this. We had been in contract negotiations for five years. We had COVID in there, so that's part of it. But we got this now just in November. I don't know if you're aware of this, but we can't just go out on strike when we want to. We have to be released from the government.
0:16:45.6 Joseph Morais: Right. Just because of the critical function.
0:16:48.0 Todd Smitala: It's very rare that that happens. I mean, you have to just jump through a lot of hoops and the President can come in at the last second and say and all that kind of stuff. But they were getting pretty close to doing that and that's what really kind of got American in line. So if we had delayed that to what's happening now, it would be a completely different situation. So it's paramount really, if we didn't get this off the ground.
0:17:13.5 Joseph Morais: Timing is everything, right?
0:17:14.8 Todd Smitala: Yeah, exactly.
0:17:16.3 Joseph Morais: So after pulling off these impressive results, I gotta ask, what's next for data streaming at APFA.
0:17:22.6 Todd Smitala: In particular as flight attendants? And we have these different schedules with our. Basically our monthly schedule comes out. It's a series of different flight sequences that each one has to fly. That information changes all the time. You know, we have delays, we have reschedules, we have cancellations. All of that information is constantly changing. And we, right now, we don't have access to real time information. With that I can definitely see, and we have kind of the proof in the pudding here of how this situation has worked, that we can set up something very similar to that in order for our members to be able to have that real time access they don't have right now. So I definitely want to be working on that.
0:18:02.5 Joseph Morais: What about stream processing? Do you think there's a future for kind of, you know, manipulating those events inside of your streams at APFA in the future?
0:18:10.6 Todd Smitala: I think so. In the near future, really? Because we do hit a little bit of that right now. I mean, we get that information coming in and then there's certain things we need to manipulate it and we're doing that right now in Azure functions because of how that had to connect with Auth0 anyway, so we're using it there. But I know we could also do it directly in Confluent Cloud. So that's something I can see. We can streamline a little bit more. Definitely.
0:18:36.0 Joseph Morais: Yeah. If you ever take a look at Flink, they have my favorite mascot. Like any tech, it's a squirrel and I love squirrels. Yeah, I'm a really big person. I could fit a squirrel in my pocket and I just thought I love squirrels and chipmunks too.
0:18:45.9 Todd Smitala: Yeah.
0:18:48.3 Joseph Morais: So you know, this wouldn't be a tech show in 2025 if I didn't ask you this question. So I'm going to do it. How are you thinking about generative AI when it comes to data streaming in your role?
0:19:00.1 Todd Smitala: In a bit more of a longer term. So if we're able to get what I want, what I'm hoping to get set up is that kind of real time access to our schedule information. If we have that, I believe that we'll be able to do some predictions.
0:19:16.7 Joseph Morais: Oh, okay. Like looking at the schedules. Let the AI look for tree, like for trends and things like that.
0:19:22.6 Todd Smitala: Exactly. Trend. Especially in terms of our reserve flight attendants that they're just sitting there, they're just sitting around in case somebody calls in sick. And we don't really know, you know, they just put a bunch of people that are available there and it always, you know, it makes people angry that they're used to sitting around doing nothing.
0:19:44.4 Todd Smitala: So I feel like if we're able to use some kind of predictive analysis, we can hopefully bring those numbers a little bit further down and a little closer to reality. You know, we know of course over holiday periods they're going to need more. But how many more? If we look at that over time, I think we can do something like that. Yeah.
0:20:01.8 Joseph Morais: See that's why I love asking this question because everybody has some initiative like something they think that gen AI can help them with or machine learning. And it's always so novel to understand how you want to use it in a particular industry. So as you were saying that like not only looking at trends from historical trends, but looking at things like, you know, maybe viral outbreaks, right. Like the flu is going crazy in this particular area. We anticipate at least two call outs or something like that just based on that and being able to feed that real time data. You know, do something like stream processing to present it to something like AI so it can track that trending and then maybe write back into streams. That would be really powerful.
0:20:37.9 Todd Smitala: Exactly. Yeah.
0:20:39.2 Joseph Morais: Yeah, that's fantastic. So, you know, let's go into our next segment. So what are the top three tools you rely on for data streaming and bonus tools that you avoid, like batch workloads?
0:20:58.4 Todd Smitala: Well, like, it's batch workloads that I avoid for sure will sell them.
0:21:01.6 Joseph Morais: I love that. Just stay away from batter.
0:21:03.4 Todd Smitala: Yeah, I mean, honestly, that's one of the worst things that we have to deal with is like, what happens if we have to, you know, batch change all of our members at one point. The system that we have set up right now, it's going to have a backlog for sure because of the speed of Auth0, really more than the Confluent side. So that's something I don't like. Yeah, I mean, some of these that we've already talked about Confluent's been amazing. You know, there's no question about that. And it's, I mean, I've, we've had no issues once it's been set up. You know, there were some issues of getting it set up, but once it's been set up, there's been no issues. Azure Functions has been good. I've learned a lot about that too, and getting that connected in there and making that work. Yeah. And that's been really good. And Auth0 has been good too. And there could be some improvements there. I wish that there were a way to do a direct sync connector from compliment into Auth0, but it's more on the Auth0 side that they're not allowing that because I've pushed them, pushed all of that on those.
0:22:07.2 Joseph Morais: Well, now, see, your voice is going to be out there through the show, so maybe other people will agree and they'll ask for the same.
0:22:12.8 Todd Smitala: Because that would make my life a lot easier. For sure.
0:22:15.7 Joseph Morais: That's fantastic. You know, so you have your data streams covered, your serverless processing and some authentication which is going to be, you know, everyone needs that, whether that's on their top three list or not. So that's a really good stack. I appreciate that. Todd, can you share a specific tactic that has significantly improved the adoption of data streaming at APFA? And I ask it because I know you guys weren't doing streaming. This is all brand new. So what did you use to kind of get that buy in. So people were like, yes, we see the value in doing this work to get away from what we were doing.
0:22:47.8 Todd Smitala: Well, I mean, I think it was just the circumstances really, because we had that issue that was like, we have to solve this. And we were under a time crunch with it and it had to be done. So there wasn't a lot of deep thought of or comparisons on how to do this. It was really just, let's go with this and we'll figure out, you know, the rest later. At this point we have to do.
0:23:12.2 Joseph Morais: This because we have to do this.
0:23:13.8 Todd Smitala: There really is no other way around it.
0:23:16.3 Joseph Morais: And you know, interestingly, in episode one, my guest, Tim Berglund, we talked about this in a different, slightly different lens. But he said, you know, finding the pain, right? And finding the pain and saying, hey, your data streaming will help this pain go away. It seems like that pain made itself very evident to you, so you kind of backed into that kind of approach anyway.
0:23:36.0 Todd Smitala: Exactly. And they can't say I didn't warn them because I said that from the very beginning. I said this is the way you should do it.
0:23:42.6 Joseph Morais: I love that. So how does APFA evaluate and select new tools for their data stack? And how are you balancing factors like scalability, cost and integration with existing systems?
0:23:54.5 Todd Smitala: Well, at this stage it's going to, well, going forward it's going to change a little bit. But from when I came on until now, we were in this phase of negotiations. So there was a much larger budget that was allowed. Now that that's done, it's really come down. So there wasn't such a limit in terms of costs. It was more like, can this do what it, what we needed to. That was really more of what they were trying to figure out and what's the most cost effective way and the quickest way to get there. That's really what we were talking about. And they really relied on me on getting the basics of that and presenting that to the leadership. You know, we talk about it with our other teams as well, but you know, they really kind of relied on what I had to say about it and would, in most cases they'd go with it. But now that we've gone through all of these issues that we've been trying to work towards for the past five years now, it's going to be time to really streamline and get the most out of our setup.
0:24:56.2 Joseph Morais: Fantastic. I love that response. So now we'll get into segment three. I don't usually, I don't always read the segment names, but I think this one's such a cool name. They have to. So our next segment is data streaming street credit. We've covered the tech tools and tactics, but let's face it, none of that matters if leadership isn't on board. So next let's discuss the art of getting leadership buy in. I know you kind of already touched on this, so I'm hoping you can expand specifically into those like key stakeholders. How did you convince leadership to get on board with your solution? Was it smooth sailing or was it a bit of a roller coaster?
0:25:31.0 Todd Smitala: There was a little bit of a roller coaster. You know, we had, I have four that I'm dealing with. I'm dealing with a treasurer, a secretary, a president and a vice president. And the one that I deal with on a day to day basis is mostly the secretary. I mean he's actually the one that's like, no, we're not going to be able to afford this and we can't do that. And you know, he's really very conscious about the cost of every little thing and he's not used to seeing the larger costs of. For me, I feel like every time I sat down with someone it was going to be $10,000 for whatever, for whatever surface it was. It was always $10,000. It was just, you know. Yeah. And so, you know, he wasn't as used to that with the treasurer. He was used to these things. He was dealing with phone systems and all of these other things and he was like, okay, yeah, you know, he could kind of figure that out. And then the president was more like, we just have to have this working and however we can get there, we've got to do that somehow.
0:26:29.0 Todd Smitala: So, you know, it was getting them all together and just making sure that the concept was going to work. And you know, a lot of it did rely on me. Am I going to be able to pull this off?
0:26:39.8 Joseph Morais: I'm glad you were confident though.
0:26:41.9 Todd Smitala: My doubts at moments. But you know, fortunately everybody's always very supportive at APFA. You know, they really are grateful for anything that we can do tech, tech wise. And when we have an improvement, everybody just, it's really a wonderful atmosphere for me to be in, I have to say.
0:26:58.9 Joseph Morais: What were the primary objections you faced? I think you did. I think it was mainly cost. But were there any other objections outside of maybe cost from a time standpoint or just from a dollar standpoint that you face from leadership or anyone on your teams?
0:27:12.9 Todd Smitala: I mean, we always have bickering on design things and you know that kind of stuff on day to day stuff, that was never a big deal. I'd say on some of the larger teams like the enterprise system, they were just like, okay, who's going to be responsible for maintaining this? And you know, how are we going to do that? Is it going to be us? Are you going to do it? You're another person, that type of a thing. Trying to figure that out. From leadership though, it was really just, you know, costs and can you get this done? Do what we need to do there. They didn't get too deep into, okay, is this version a better idea than this version? You know what I mean? Could we do this? Could we? What is an alternative to this? We didn't really get into too many of those.
0:27:57.2 Joseph Morais: Yeah, it makes sense. Like when I think about your journey through Buy in, it was really about hearing the requirements of each of those different kind of stakeholders. And they were different. Someone's like, hey, we need this to be cheap. Hey, we need this to work. And I think by, you know, addressing each of those individually, even down to something I think is very important, ownership, especially at post implementation, which I think a lot of people forget, like, someone's got to run this system. It's up and running. By addressing all of those concerns individually, it sounds like over time you got all the buy in you needed and obviously the results speak for themselves.
0:28:31.4 Todd Smitala: For sure.
0:28:38.6 Joseph Morais: We've cracked open your playbook, we've unpacked tools and tactics and we've explored your real world win in building with data streaming. Now let's shift gears and dive in something extremely hard hitting our data streaming meme of the week. What we need to do now is get focused and stop pointing fingers. You're a problem. You're a real problem. Yeah, like that was me thinking about you talking to Frankenstein. Like that was 100 where my mind went when I heard that meme.
0:29:08.3 Todd Smitala: Yep, Exactly. Yeah, there were a lot of, you're a problem. You're a real problem. And it was coming to me saying, yeah, why isn't this working?
0:29:20.1 Joseph Morais: You know, and you know, this is the reason I love this meme, not only because I thought it was extremely appropriate with the conversation and your title and everything, is that this is a problem. I think almost every enterprise has. Right. Unless you're like digital native and like everything's a startup and you kind of maybe, you know, figured out all your tech stack in the last 10 years. But that's not the reality for most companies. Most companies have been around for a long time. They might have things like mainframes. If they were around long enough, they may have things like Vax systems and they probably have some really bad pipelines right through legacy tech, a lot of batches, things on crime, jobs and man, once you start to build something new and you know you have that reality like you had in your relationship with Frankenstein, it's such a hard stop to innovation. Right. Like once your data's flowing, there's so many things you can do.
0:30:12.2 Todd Smitala: Right. I mean from when I came on I was like, we need to scrap all of this and start from zero. We need, we could even have a Ruby on Rail situation that would have resolved everything that we needed to do. Like just, you know, had the server, we could have built all of that stuff. But of course we need a team to do that and you know, you have to deal with what you're starting with and then try to make those, implement those improvements over time and let's see what we can get.
0:30:43.4 Joseph Morais: Before we let you go. We're going to do a lightning round. So byte sized questions, byte size answers and that is B-Y-T-E. Sort of like hot takes but schema backed and serialized. Are you ready?
0:30:53.4 Todd Smitala: I'm ready.
0:30:54.9 Joseph Morais: Excellent. What's something you hate about it?
0:31:00.4 Todd Smitala: Upgrades. And when one thing pictures and 10 things break. Yeah, when you, I mean, you know.
0:31:08.3 Joseph Morais: You know you have to do it.
0:31:11.1 Todd Smitala: I've been living with that.
0:31:13.2 Joseph Morais: Yeah, that's a great response. It's the first time I've gotten that one. But it's. I think we all fear that upgrade. Like what happens after the thing reboots. Right. What's the last piece of media you streamed?
0:31:23.1 Todd Smitala: Oh, I just was watching a visual album from a band called Miana System. They just launched it. So. Interesting. Yeah, it's kind of like a video.
0:31:34.7 Joseph Morais: So what's that, like a YouTube video? That is the full album?
0:31:37.6 Todd Smitala: Yeah. The whole album though, just kind of. It's a story, but it goes through each track and brings it to the end. It just launched day before yesterday, so yeah, it's really cool.
0:31:47.3 Joseph Morais: That's awesome. If Pink Floyd were a modern band, I'd imagine all their albums would be visual.
0:31:51.3 Todd Smitala: Yeah, for sure.
0:31:53.1 Joseph Morais: So what's a hobby you enjoy that helps you think differently about working with Data across a large enterprise.
0:32:00.5 Todd Smitala: I'm out on the beach and out kayaking as much as I can, and then it just clears the mind and then I come back refreshed and I'm ready to think about things in a different way. For sure.
0:32:10.3 Joseph Morais: So it's more about clearing the mind as opposed to doing something that kind of says, oh, we should do it that way. It's just. It's all about hitting that Zen state.
0:32:19.0 Todd Smitala: Yeah. Although I am a musician as well, and so there are those moments when I'm in the middle of something and it does, you know, it comes to me like, oh, we could do it that way. While I'm in the middle of... I play piano, obviously.
0:32:32.4 Joseph Morais: I love that.
0:32:33.1 Todd Smitala: It will come to me. You're right. I wasn't thinking of it in those terms, but yeah.
0:32:38.4 Joseph Morais: No, that's good. And I mean, I'm the same way. And anytime I hit a point of frustration and this isn't just limited to my career. This is almost anything in life. I like to walk away from it. Like, I like to completely remove myself from it and then come back with a clear mind. Because once, like, you start getting emotional or frustrated, It makes it impossible to see clearly. So you getting out on the beach and clearing all up, that's a... I got to start doing that more often. Can be a little tough because I'm in Arizona. Can you name a book or resource that has influenced your approach to building event driven architecture or implementing data streaming?
0:33:11.4 Todd Smitala: Honestly, it was the library at Confluence. I mean, between all of the documentation and, you know, there were a ton, really a lot of videos. And in courses, you know, I did a good amount of them as I was, you know, looking through this whole option, you know, this whole possibility of using this and. Yeah, I mean, everything from setting up the, you know, the Klee and all of that kind of stuff, it was really interesting. I learned a lot that way. I didn't really go to any other vaults. I just kind of followed all the information that you had there.
0:33:43.3 Joseph Morais: Nice. Well, in that case, I'll plug box docs Confluent.io and developer. Confluent.io is where you found all that great information.
0:33:50.4 Todd Smitala: Yeah, and when I first started with Confluent, it was really. There wasn't a lot of support. You know, it was just me trying to figure things out. So I didn't have somebody there yet to go to and see not until we kind of signed up for an enterprise system, so.
0:34:05.2 Joseph Morais: Well, I'm glad you were able to do, you know, evaluate the system without actually, you know, having to make an initial purchase. I know we do a lot of things that allow you to run things locally in Docker and kind of get your feet wet and understand it also I'm really thrilled to hear that that's useful for you because you know, we all use it here and we try to you know, make you know, sent these links out but it's really nice to hear that people are getting value out of that. So what's your advice to a first time chief data officer or somebody with an equivalent impressive title?
0:34:33.4 Todd Smitala: Equivalent. I love it. I would say don't, you know, don't give up. You know, something may seem daunting. It seemed very daunting when they first raised the issue of Kafka for me and nobody on any of my teams seemed to know where to go with that. And as I said, I just kind of dove in and started reading more and learning more and teaching myself a bit. You know, don't stop with that. Always keep learning and always keep going with things and it will eventually work out. You're going to find your way.
0:35:00.7 Joseph Morais: I love it. Todd, any final thoughts or anything to plug?
0:35:04.0 Todd Smitala: I would say that I'm really excited about the future with Confluent because there is a lot if we're able to do my next plan, my next evil plan I'll say of getting that real time sequence data. It's going to be a game changer for flight attendants at American Airlines for sure. So I'm very excited about that.
0:35:23.5 Joseph Morais: I hope you get that done and it goes well because I really do, I agree with you. That will be huge for all of your members. So thank you so much for joining me today Todd and for the audience please stick around because after this I'm giving you my top three takeaways in two minutes. Man, that was a great conversation with Todd. Here are my three top takeaways from today's episode. Data streaming ensures all voices are heard. Right. So what they built at APFA on top of data streaming ultimately allowed those 27,000 members voices to be heard even louder than previously. So going from around 50% up to 94% or over 94% participation rates, just absolutely incredible. And who I'm sure most of you have already been on flights before hopefully in the United States and, or anywhere you've been. But you understand how much care and safety goes into flights and the flight attendants being those shepherds of our safety and our and our comfort and ensuring their voices get heard ultimately probably to make the experience for us all better is really fantastic. I love Todd talking about Frankenstein's right or some people might call that dirty data integration.
0:36:34.3 Joseph Morais: He called it a Frankenstein. I love it. And the truth is even with modern architecture sometimes you're going to have things that don't work exactly the way you want, but you want to keep those Frankenstein small. And the last thing I want to give you as a takeaway. His advice to a Chief Data Officer: don't give up if you know that event driven architecture, data streaming are the right way to go forward. Get those positives out there, ensure you get buy-in from everybody you know. Keep doubling down because it's very important to express why we're doing something because change is not easy and going from something legacy into something new is very hard. But once you convince people of the benefits on the other side of it, hopefully you can get their buy in. So do not give up. That's it for this episode of Life Is But a Stream. Thanks again to Todd for joining us and thanks to you for tuning in. As always, we're brought to you by Confluent. The Confluent data streaming platform is the data advantage every organization needs to innovate today and win tomorrow. Your unified solution to stream, connect, process and govern data starts @confluent.io and if you'd like to connect, find me on LinkedIn. Tell a friend or co worker about us and subscribe to the show so you never miss an episode. We'll see you next time.