Nnamdi Iregbulem, Lightspeed Venture Partner

Thank you for tuning in to today’s episode of the Breaking Changes podcast. I’m your host and Chief Evangelist for Postman, Kin Lane. With Breaking Changes we are exploring the world of APIs through the lens of business and engineering leadership. Joining me today we have Nnamdi Iregbulem, partner at Lightspeed Venture Partners. Nnamdi has his finger on the pulse of what’s happening when it comes to data and real-time infrastructure, and he shared with me a little bit about how he got here, but also a glimpse at what is hot from an investor point of view. Let’s just dive right in.

Who are you, and what do you do?

Awesome, well, thank you, thanks so much for inviting me to do this. My name is Nnamdi Iregbulem, I’m a partner at Lightspeed Venture Partners, which is a long-running venture capital firm, where I focus on early stage software investments in all things technical. So developer tools, application infrastructure, data science, machine learning, basically anything that serves a more technical audience, primarily developers and data scientists. I’ve personally been a self-taught programmer for a long time, so it’s really fun to put my money where my mouth is.

Yeah, so why investment over programming, what made you go down this road?

Yeah, it’s a good question. So as a backstory, I’m the firstborn son of two Nigerian immigrants who really wanted me to be a doctor someday. And in some ways my initial forays into programming were trying to find some other thing that I could do. And I fell in love with it, tinkering with programs, building websites, modifying games that I was playing, building computers, and it being pretty obvious to me that this was the route that I wanted to go. But the other kind of residue of having immigrant parents is that there’s sort of the sense that you are what you study. And I wasn’t sure I wanted to be a software developer, so I didn’t want to major in engineering. And what I did find as a great study area was economics. I saw what was going on over there and all the different positive and negative effects that finance can have on the world, and I thought that was pretty interesting, so I decided to study that. And I was kind of on that route for the most part, but it’s always with the focus on technical tools and technology generally.

Yeah, very interesting. I like your, I would say investment is not a route I would have gone in my career, but I really appreciate technical folks choosing to go, because it really showed in our brainstorming conversation, the depth that you had. From other investor conversations I’ve had, you definitely have your finger on the pulse of what’s going on. And for me, I’m in the API space, I come from a database and technical background, so I come from the technical part, but I quickly saw that there was more to all of this than just the technology, that the business of APIs really mattered, doing APIs as a product, but the politics and regulation. As an investor, there’s a lot more to it that makes a product, a company successful, or a technology or a platform successful. And then I would say, as I work at Postman, Chief Evangelist at Postman, and we’re a dev tool company, so you kind of opened the door there with dev tools, but you rapidly went into some other pretty advanced areas.

So let’s start with dev tools. Why do dev tools matter in 2022 when it comes to investment? Has it all been done, or is there interesting things happening right now?

It definitely hasn’t all been done, there’s almost too many interesting things happening these days. But from a first principles standpoint, the reason I care so much about developer tools is because I care so much about developer productivity. I think that productivity in a modern knowledge-based economy is the way that the typical person kind of advances themselves, earns higher wages over time, contributes more to society over time. And so anything that can help enhance the productivity of the average knowledge worker is really interesting to me. And it’s especially interesting to me in the context of software development, because there’s always these shiny new tools, basically every day it’s not one or another coming out, and oftentimes it’s all really interesting, but sometimes productivity is sort of an afterthought, and it’s not always obvious that this new thing is actually moving the ball forward from a productivity standpoint. And so I kind of try to use it as an important lens in everything that I do. I also think it’s important from the perspective of, if I want more and more people to be able to become software developers, and I want that talent to grow over time, for companies to onboard more engineers over time, they need to see the gains from that, the ROI, and that comes immediately in the form of productivity. And so if I can help people ship more software faster, then more people can become engineers, more companies will want to hire those folks, and it just kind of creates this really nice flywheel effect almost in the ecosystem. So developer tools are kind of at the core of that. I’ve been a big investor in the category for a while, and probably my claim to fame would be sort of investing in GitLab, which went public recently. But always trying to find the next one too, but don’t tell Sid that.

Yeah, GitLab made a pretty big splash, I would say. I didn’t really see it, I would say in the shadow of GitHub at the moment, because I was kind of captivated by what GitHub did with Git, and making Git more social and more modernized. But GitLab, I would say, really surpassed in my mind as far as actually hitting on the productivity piece that you talked about. It wasn’t just about public repos and public open source solutions, it really hit on what enterprises need and are really needing to enable that productivity, and not just from a top down but from a bottom up too, like actually making developers happy in their day and successful. And that’s what I heard in your definition of productivity, is it’s not just productivity for the sake of business productivity, it’s productivity so developers are happier, healthier, there’s a wider pipeline of people striving to do things. So I like that, I support that, so I’d love to help keep sharing notes as far as dev tools out there that I think is the next GitLab. But I don’t have any off the top of my head.

Oh yeah, yeah. So let’s rattle through some of the areas. So service mesh has been one area that investors have been talking about for a while as being the next thing, and there’s definitely some things there. But I’m really interested at the gateway layer, and what are you seeing next when it comes to proxies and gateways, ingress, egress, like what enterprises are needing, because this is the API layer for me. So is it real-time infrastructure, is it more service mesh in that circuit breaker type stuff? What are you seeing at that dimension that’s moving and shaking things?

Yeah, there’s a couple different things I could talk about. One of my favorites is something you mentioned, which is around real-time infrastructure and the shift from, let’s call it the batch paradigms of old, where you had a database or a data warehouse of some sort, it would be updated on a more infrequent basis, maybe once a day or even once in a longer time period. And so whenever you query that database, you would basically have stale data at any particular moment, unless you were just so lucky as to query it right after the update. And as a result, your businesses weren’t able to adapt in real time to data assets changing in their business. And so execs and whoever else cares about that data can only see it’s a week ago of information, or two days ago, and things could be out of date by then, or irrelevant at worst. And so shifting from batch to a higher frequency of data output requires a bunch of changes in the underlying architectures, it’s not as simple as just turning a knob. And so there’s been this emerging category of technologies that help support this, and I’ve been lucky enough to be an investor in a few of them. There’s sort of the whole streaming ecosystem, which Confluent has done a great job evangelizing via Kafka. I’m an investor in a company called Redpanda, which is sort of an extra take on streaming data, built from scratch, a streaming engine built in C++, which is highly performant and also actually really easy to use. Another example is a company like Materialize, which is basically, let’s say you have a stream of data coming in and you want to create a materialized view of that data that’s constantly updating as the data is coming in and changing, that’s actually technically quite difficult to do. But the team there built this technology called timely dataflow, that’s actually able to do what you kind of call change data capture as data is coming in, update the database in real time, enable you to query at very low latencies and get very fresh data, which is like the classic trade-off that you typically face in most systems. It’s either you get low latency or you get really fresh data. Historically it’s been very hard to get both, and so Materialize helps you kind of break that paradigm. We’re also investors in a company called ClickHouse, doing stuff around real-time OLAP, online analytical processing, and a couple other things that are coming down the pipe soon that I can’t talk about yet. But really just a big fan of real-time data infrastructure, and I think now at this moment, so that’s one piece that I think is really interesting.

Another thing that’s kind of interesting, for folks who might be on the line who are involved in running cloud-hosted managed offerings for universal infrastructure solutions, is this idea of peering from the on-prem to the cloud. And on-prem could be defined as not necessarily literally on-prem, but be self-managed or what have you, to some kind of managed offering. And we’re officially entering a world where hybrid cloud is kind of the steady state scenario, where you have some stuff that you’re managing yourself and you have some stuff that’s being managed by some vendor in the cloud. How do you get these systems to talk to each other in a secure way, oftentimes with traffic going over the public internet, which is a big red flag for a lot of CSOs and security-focused folks? And so there’s an emerging set of different tools and technologies that one can use to build that bridge. So you could have an open source version of some infrastructure technology, like Redpanda or like Elastic, your particular pick, run it in your own virtual private cloud, also have that connect to the managed service running from the vendor itself, like Elastic Cloud or Redpanda Cloud, and do that in a way that doesn’t raise any kind of red flags or security issues. I think that’s really interesting, I think it’s going to be increasingly a paradigm for these infrastructure vendors. So that’s another thing that I’m very closely tracking.

Yeah, it really sounds like, I mean, REST APIs or simple web APIs has been kind of scratching away at the legacy database infrastructure. And I would say the gateway has, it’s got a long way to go, but it’s starting to shift the power center from the classic database, which is the world I come from, 80s and 90s database, the power center of any enterprise organization, and those guys can write their own ticket, they have jobs forever, they have a lot of power and control. But the gateway has started kind of pulling that away. But this real-time shift, and I agree, what I’m seeing as well, and event-driven laying on there, is how do we make it more real-time, how do we map it to our real-world needs when it comes to, well, we have these existing legacy databases, we have some things that are on-prem, some things are in the cloud or in managed services, how do we map to that, but then how do we make it not just request and response, so synchronous, but asynchronous, so that we can get notified when events change, we have dashboards that are always up to date. And so what are you seeing at this layer when it comes to data science or data ops? You talked about dev tools and empowering developers as part of the microservices evolution and this shift to the gateway power, DevOps too, developers have a lot of power now, they’re really important to a lot of operations. So when it comes to data science, data ops, what are you seeing there?

Yeah, it’s really interesting. There are a number of different paradigms in software development that people have tried, to varying degrees of success, to apply to the data ecosystem and data science. The notion of DevOps, agile development, Git-based workflows, etc. And so there’s an emerging set of tools that you could call either MLOps in this machine learning specific area, and then this modern data stack, which centers around typically a cloud data warehouse, the source of truth for data in your organization, and then a bunch of different tools that either feed into that or feed off of that. As I mentioned, some software development paradigms transfer well, some don’t. And what’s interesting is that, especially in the data science world, the skill sets that make you a great software engineer are not perfectly overlapping with the skill sets that make you a good data scientist. And oftentimes folks who are brilliant data scientists are just terrible at traditional software development, and they’ve never shipped anything into production. And so there’s this very natural friction point that occurs when, okay, if you have a machine learning project of some sort, some data science thing, it kind of works in the prototyping phase and you can test it and it looks okay, but then when you want to put it in production and you have to do all the things that harden it to do that, it’s like a real barrier for many organizations. It’s like that phrase, crossing the chasm when it comes to data science. And so I continue to think that’s a big gap most organizations have, and I spend a lot of time looking at different tools to help people bridge that gap. And there’s a lot of interesting things happening around the modern data stack. I mentioned folks like Snowflake, Redshift, BigQuery, these cloud data warehouses that are extremely cloud-native, you have folks like Fivetran that can feed data into those data warehouses, and then folks like dbt or Matillion that extract data out, or I should say transform the data in the data warehouse, which is an important element. I’m really excited by this ecosystem. It’s been this real big shift in just the past few years. Like five years ago I don’t think I had heard of dbt, and now it’s like you can’t avoid hearing about it. I’ve been following the Snowflake story for longer, but still it’s really been a shift over just the past few years. And so it’s all really exciting. I’m curious to see where it settles in terms of paradigms that work for data and data science, how much those end up being similar to or different than the paradigms that work for modern software development.

Yeah, the pool, data scientists, going to talk about increasing the pipeline of developers, it’s not easy to produce data scientists. So do you see tooling and startups as being how we’re going to augment them with exoskeletons to have the superpowers they need? Or maybe data ops or DevOps people to have more data scientist-like skills, is that possible through this tooling?

I think it is, especially if you take some of the ML and data science tooling and integrate it really nicely with this modern data stack. There are a number of folks who are attempting to do this. There’s a company called Continual that’s basically trying to create very much a modern data stack native approach to ML. It makes it quite easy for even a less technical person, just like a data analyst who works with data but maybe doesn’t know how to write code and isn’t that kind of person, to connect to your data warehouse, train and deploy a state-of-the-art machine learning model, and then feed those predictions back to the data warehouse as a source of truth. I think that’s really interesting and could be the right form factor to help solve this, which was basically a labor supply shortage in this ecosystem. And there’s more that’s being worked on in this area, but I think it’s all really interesting. And I think you might be right that there’s especially more of a labor supply shortage in data science than there even is in software engineering, which is saying a lot.

I would say this doesn’t sound like our grandfather’s AI or ML. It doesn’t feel like it’s Watson anymore, the omniscient one you train, you just give all your data to this one omniscient, god-like presence, it knows everything and you ask these questions. This type of machine learning seems much more granular, atomic. Is that what, what is MLOps and ModelOps, I keep hearing about how this is supposed to be enabling this realm, what are those?

Yeah, it’s sort of like creating your machine learning models very similar to the way you would treat, maybe not an application, but a service almost. It’s one way of thinking about it. And think about these more like, as you were putting it, like atomic ML models that are really just, at the end of the day, functions. A model takes in an input and spits out an output, and so you think about that just like a function, and you can run it just like any other function you write, like a microservice, you can run it like a serverless function of some sort. And so the set of tooling that’s emerged for how do you manage all that and wrangle all that any organization is going to have, but models running at the same time, there’s not like one master model, as you were saying. And so it almost ends up being like, you have a microservices fleet in your software development, you have a model fleet now. And so it requires some sort of tooling just to rationalize it. So yeah, I think that stuff is really interesting. There’s so many vendors in the space, we’ll see which is going to actually create a lot of value. It’s like you look at these MLOps landscapes and there’s a million and one logos on there, but folks are working on it.

Yeah, I’m always curious, it’s tough to keep up with the startups and understanding what’s actually making people’s lives, on the ground floor. But it sounds like this machine learning’s got to be much more real time, that those models, whatever the solution we choose, our models have to iterate and evolve along with this real-time data as well. We can’t just be constantly training and doing that work, it’s got to get automated. Or are there, like, Git flows for machine learning, how’s this managed?

Yeah, it depends a little bit on what approach you take from a modeling standpoint, but as a general rule these models can be framed in terms of code that you could put in a Git repository, or in terms of a data set that you can store like anything else. So that’s the approach that folks typically take, and then to your point around version control and whatnot. And then to your point around real time, and can’t talk about it yet, but have some stuff coming down the pipe in this area, you kind of hit the nail on the head. Things are moving real time, a lot of models are actually not meant to be operated in real time, they become modified in a certain way that’s its own skill set. You have a model that ingests data, outputs predictions, has to be modified, and so that’s a very underserved category today, but a very fast growing one in terms of needs from businesses. And so, excited to support that category, and more to come in the near future.

Yeah, I look forward to that, because I could see testing emerging, like you’ve got to test these models, because I’ve seen modern ML be compared to puppies and kittens. It’s like, it grows and it changes, and sometimes it gets all Frankenstein on you, even at a small granular level. So you got to be testing, well, wait, that batch of real-time data did something to the kitten that we’re not cool with.

So when it comes to this real-time infrastructure, the customers who are buying these services, are these large enterprises, is it scope of operations that drive people to this? Because I’m getting a lot of questions from API people who have the last five years of microservices and they’ve dabbled in GraphQL, now they’ve got REST APIs, they’re building them as products, but they’re like asking me, you know, what’s next, what should we be investing in? And what’s that trigger? Is it size of data, scope of data, demand and volume of requests, what drives that real-time growth when it comes to the enterprise?

Yeah, I think there’s maybe two emerging constituents interested in real-time infrastructure. One are folks who have an extremely high frequency or cardinality of data. These are typically folks in the consumer arena, where you have hundreds of thousands or millions of customers, and those customers are interacting with your product, which generates a bunch of real-time clickstream data. Your product is oftentimes making recommendations to those folks, which needs to be generated real time based off their past behavior. And there are various ways you can in real time prompt the user in ways that might increase conversion rates or increase how much they order. And so this set of, whether it’s e-commerce, marketplace businesses, various consumer apps, they just have a natural tendency to operate in real time if possible, and they’ve always been pushing the edge in terms of the infrastructure, just due to the sheer volume of data that they generate. So that’s the one category that I’ve seen emerging. Then another constituency is basically large, maybe older enterprises that are doing various things around digital transformation, and one of those pillars under that is, how can we move faster, how can we not be this lumbering giant that sees a problem and then five, six months later is able to react to it, or even longer, how to react in a shorter cadence and decrease decision-making and cycle times. And so real-time data is a way to do that, where you can almost create these applications that on their own react to data as it’s changing, and so you need these systems to support that. There are other folks who are going to come up in this space as needing this kind of technology, but those are probably the ones that I’ve seen most commonly so far.

Yeah, I’m always curious, the companies who come to me asking these questions, where they’re at in their overall journey matters a lot, where they’re coming from, how they’re seeing things, and how heavy they’re putting their foot on the gas pedal. What are you seeing when it comes to observability at these layers? Because that’s a big one, I know folks are struggling with just the massive growth in data and real-time activities, being able to observe and not just understand what’s happening and see it in a dashboard but be able to respond to it. And hearing the real-time nature, the urgency there, what are you seeing from an observability vantage point?

Yeah, so observability is just a massive market, no matter how you look at it, whether you look at application observability, data observability, model ML observability is emerging as its own category. Similarly, when you have these very large data sets with log data, metrics data, or data that has very high cardinality or complexity, you end up needing pretty sophisticated observability tooling just to understand what’s going on, filter down to what matters amongst all the things going on, there’s so much noise for every unit of signal. There’s so many things that can go wrong in these complex systems. And so there’s a bunch of different tooling that’s emerged around this. We’re invested in a company called Grafana, which is doing some interesting things in this area, to put it mildly. We’re invested in another company called Polar Signals, which is what’s called continuous profiling, which is leveraging some of the interesting things happening in the Linux kernel via eBPF to better analyze what’s going on from a performance standpoint. And then there’s this emerging set of data observability tools you could call them, that are also benefiting from this shift to the modern data stack, and those are actually quite interesting. I could go on, but it really is a big market, there’s lots of things to track down. And the last thing I’ll say is, it’s a new market in the sense that there’s been tools for trying to figure out what’s going on in your infrastructure, in the security space and the general log storage space, like Splunk has been a player for a long time, but it’s also kind of quite expensive, and folks are looking for alternative options at this point. And so things that can provide the same level of functionality at a cheaper price, oftentimes via open source, are gaining a lot of steam these days.

Yeah, I’m always fascinated by how we see all of this, because it’s so massive. And I have access to a lot of data just from, we have 18 million developers with Postman, and trying to see that and understand and see it in ways that matter, and then see the health and state of things, not just the data that’s being generated but how well things are performing and what direction we’re moving, all of that is really hard. And it’s perpetual too, it’s not like, oh, now we can see everything, we’re good to go. It changes next week, and we’re having to do things.

From your vantage point, do you have any views into regulatory influence at any of your startups or investments? Is European regulation or US regulation playing a role in what you’re seeing?

Yeah, I think right now the big focus is around privacy, and whether it’s GDPR of Europe or CCPA in California, which even if you aren’t based in California, you end up kind of needing to adapt to, it sets the standard for much of the country. Teams have to be very careful around not breaking any of these laws, you build your super-duper awesome data infrastructure technology, and there’s an emerging set of technologies that are trying to make the anonymization of data, removing of personally identifiable information from data, make that process easier and easier. These tools have existed historically, but they’ve been a little bit of a science project in terms of just their complexity, and how you deploy this thing and what is it actually doing. And then from a data quality standpoint, when you privatize this data, oftentimes you reduce the data quality to the point of being kind of useless, and so how do you balance the quality of data with the privacy demands that the regulators are throwing at you? I think it’s an important thing that should still be kind of worked out in real time, I would say. But that’s the main regulatory angle.

The only other one that I would keep a close tab on for now is, I think it really has a term associated with it, but it’s basically this idea that if you get hacked, you’re liable. And so you don’t get hacked, and increasingly I think we’re seeing that there’s no way to actually prevent a hack in the long run. Eventually your systems will be breached, there’s too many holes, the attack surface is too large to avoid that. And so how can you be most resilient to it, conditional on it probably happening? How do you reduce the scope of data that a breacher or attacker might get access to if they do manage to get in? How do you reduce the damage that they’re able to do? How do you prevent them from these ransomware attacks that people are doing, when they get into your data and then lock it down and take you hostage basically, how do you avoid these really bad tail risk scenarios? I think it’s an increasing area of focus, and again, I don’t know if there’s a regulation in particular that will come down around this, but if there’s massive legal liability for companies, this is like what you want to avoid if you’re a CSO or CIO, for sure.

Yeah, I spent last night reading a cybersecurity from the cybersecurity agency in the US, talking about the cost of a data breach and how they figured out, but basically it comes down to scope, as you said, the number of records, the type of record, is it PCI, payment information, is it personal information, or is it health information? Because health is the number one, because of PHI, and then how long until things got back to normal, so how many days did you sit on this before you responded? And those were the main dimensions of the cost. There was like sliders that you could go, okay, we waited six months and it was PHI, oh, it cost you millions. And so it’s an interesting realm, and I don’t know of any regulation, but I think the teeth are getting put on. I think a lot of breaches, companies were able to know cost of doing business, we shelled out some money, it hit us in the press, it wasn’t too hard, but I think the damage is getting greater, but also the teeth on regulation, I think we’re going to see a lot more of that coming. So it’ll happen, it’s just a matter of when and at what scope, regionally, countrywide, statewide, that kind of thing. But like you said, with CCPA, California can set the bar, set the tone, Europe can set the bar, set the tone. So we just got to pay attention to these things.

Yeah, from the investment side, I’m always learning and I’m trying to understand the investment side. So you invest in early stage through like B and C rounds, is that because financially that’s the sweet spot, and/or because you have more control over the direction and helping make them more successful? Why is that the sweet spot for what you guys have chosen to invest in?

Yeah, so it’s interesting, I’ve been on a little bit of a journey on this point myself, actually. When I started off as an investor, I was more of a what’s called a growth stage investor, so I was mostly in a series C and later. And since joining Lightspeed I’ve been more focused on the early stage, as you said, kind of seed stage through series C, or even series B. And I think both stages are interesting in their own right. What I find fascinating about the early stage is that there’s just so many, the vector that shoots outwards could take many different paths from that point, and being able to have an influence on that, being able to go on that journey with the founder, is really exciting to me. I’ve always been a bit of a product and tech nerd, and so at the earliest stages is when there’s still real product and technical risk, and being able to understand that as an investor can really pay off. And so I find that super rewarding. I would be nerding out about this stuff whether or not I was involved in the businesses, but now I can do that and it’s part of the job, so that’s always great. And being able to go on these journeys with founders, as I mentioned, it’s just a really special thing. You get to see them evolve, you get to see the company evolve. It’s kind of a blessing to do that without having to, frankly, have much personal risk yourself, which is always kind of a weird dynamic, I think, in Silicon Valley between VCs and founders, which is probably a longer discussion. But that’s what I really like about the early stage. You need to have an opinion, you need to have a viewpoint, you need to have a perspective. And I’m an opinionated person, and I like making bets on really technically brilliant people building awesome technical products, so it’s been a good match.

Yeah, I mean, I’m a developer and I’ve built, I would say, about 14, 15 startups, I’ve only had one that was successful. I’m working on, you know, Postman, I’m not a founder in it, but definitely invested in it being successful. But early on, I’m just fascinated over the last decade what I’ve learned about investing that I didn’t know, just how nuanced and how much it’s needed for successful products and services to get out there in front of people. So I’m interested, I like your approach, I like your style to it, and what you guys are focusing on, definitely technically it caught my attention, but the fact that you’re investing at this front line of the API realm and data, for me being the backbone or the blood supply and all of that, is really interesting.

What do you do beyond investing in tech? What do you do to keep yourself an interesting person, happy, well-adjusted individual?

Yeah, so there’s some things that are closer to the stuff I do at work and there’s some things that are further. In terms of closer, I’m a big nerd both in and out of work, and I majored in economics in undergrad, still read a ton of economics. I’ve actually learned more about economics since graduating than I did actually in college. So I read probably two or three academic economics papers each week, and I just find it really fascinating. The analysis is actually very analogous to what people do in the machine learning world, and so there’s a lot of cross-pollination of interesting concepts that apply in both areas, so that’s really awesome. And then some computer science papers as well, especially in the machine learning area. And then more further away, I’m a really big dancer. Dancing is like my favorite thing to do, I’ve also been mostly self-taught there. And ever since I was a kid, I remember I played football in high school, and my favorite game each year would be the homecoming game, and I would just cross my fingers before the game started that I wouldn’t get injured so that I could dance at the homecoming dance that night, which were my favorite things. In college I did ballroom dancing, my freshman year roommate was really into it and got me into it, and it was one of the more amazing experiences I had, and I’ve kept up with dancing since. So that’s my favorite thing to do, something that surprises people given how I spend my time otherwise, but it’s awesome.

No, it’s whatever you got to do to nourish your soul. And I feel in the tech space you’ve got to really invest in those things. And I would say dancing is always associated, for me, with live music. I used to have record labels and tour companies in the 1990s because I was good at building databases to support their operations, and so being at live shows, like behind the mixing board, usually backstage or something, I started dancing and doing, I really found it was enjoyable, and it’s something I think that has disappeared from my life that I should probably bring back, because I feel like I don’t have anything like that, and that’s not tech, something that’s just so about your body and being there and being present. So I like it.

Yeah. Well, I really appreciate your time today, this has been a great conversation, interested to learn from you. The real-time infrastructure I think is a big part of what I’m trying to educate and understand right now, so it really helped me there. Part of my team is the AsyncAPI team, and they’re really focused on defining event-driven APIs and that realm, so I learned a lot. I got to dive more into the data science and the MLOps part of it, observability, traceability, these are all things that are on my list to learn. So I got you in my Rolodex, I would love to have you back, I think we can maybe explore some other possibilities down the road. But until then, I appreciate your time today. Thank you so much, this is great, I really appreciate it.

Thanks again to Nnamdi for stopping by. For more on Nnamdi, you can find them on Twitter as Who Is Nnamdi, and whoisnnamdi.com. You can subscribe to the Breaking Changes podcast at postman.com/events/breaking-changes. I’m your host, Kin Lane, and until next time, cheers.