The conversation this week is with Brandon Satrom. Brandon is the VP of Developer Experience and Engineering at Blues Wireless, a driven technologist and experienced leader with a background in product management, strategy, architecture, software development and developer advocacy. He describes himself as a technologist first, and loves to use what he knows and learns to teach others how to build things and solve problems with the latest technologies and platforms. Finally, he has a mentor at the RIoT Accelerator Program, and will actually be presenting at our applied AI meetup on July 7, he'll be speaking on Smarter Everything with the ML and the IoT.
If you are interested in learning about how AI is being applied across multiple industries, be sure to join us at a future AppliedAI Monthly meetup and help support us so we can make future Emerging Technologies North non-profit events!
Resources and Topics Mentioned in this Episode
Enjoy!
Your host,
Justin Grammens
The conversation this week is with Brandon Satrom. Brandon is the VP of Developer Experience and Engineering at Blues Wireless, a driven technologist and experienced leader with a background in product management, strategy, architecture, software development and developer advocacy. He describes himself as a technologist first, and loves to use what he knows and learns to teach others how to build things and solve problems with the latest technologies and platforms. Finally, he has a mentor at the RIoT Accelerator Program, and will actually be presenting at our applied AI meetup on July 7, he'll be speaking on Smarter Everything with the ML and the IoT.
If you are interested in learning about how AI is being applied across multiple industries, be sure to join us at a future AppliedAI Monthly meetup and help support us so we can make future Emerging Technologies North non-profit events!
Resources and Topics Mentioned in this Episode
Enjoy!
Your host,
Justin Grammens
Brandon Satrom 0:00
You know, as with any form of thing that falls under that umbrella of what we like to refer to as digital transformation, I think a lot of companies when they're looking at these things, tend to start by thinking in terms of new revenue opportunities, new lines of business and things like that. And that's useful. But I really think that there's a whole lot of opportunity under this idea of cost savings and helping companies actually run their businesses more efficiently.
AI Announcer 0:25
Welcome to the conversations on Applied AI podcast where Justin Grammens and the team at emerging technologies North talk with experts in the fields of artificial intelligence and deep learning. In each episode, we cut through the hype and dive into how these technologies are being applied to real-world problems today. We hope that you find this episode educational and applicable to your industry and connect with us to learn more about our organization at applied ai.mn. Enjoy.
Justin Grammens 0:56
Welcome everyone to the conversations on applied AI Podcast. Today we're talking with Brandon Satrom. Brandon is the VP of Developer Experience and Engineering at Blues Wireless, a driven technologist and experienced leader with a background in product management, strategy, architecture, software development and developer advocacy. He describes himself as a technologist first, and loves to use what he knows and learns to teach others how to build things and solve problems with the latest technologies and platforms. Finally, he has a mentor at the R!oT Accelerator Program, and will actually be presenting at our applied AI meetup on July 7, he'll be speaking on Smarter Everything with the ML and the IoT. It's awesome that you're mentoring giving back, Brandon, this is something that I have huge respect for. So thank you for doing all that you do to share your knowledge with the community. Welcome to the program.
Brandon Satrom 1:41
Thanks, Justin. I'm happy to be here.
Justin Grammens 1:43
Well, great. You know, I gave a little bit of background with regards to what you're doing today. But maybe you could share with the listeners, you know how your career has progressed, I guess you know, where you started, and how you got to where you are today.
Brandon Satrom 1:52
Yeah, great. So I have been working professionally in this technology space for 22 years now started, you know, way back when in the late 90s. Doing primarily client server development and the Windows world spent several years as a front end engineer, web developer in the early 2000s. My career kind of has two halves. There's a half of the career that I spent mostly in the front end world working for, you know, large fortune 500 companies doing consulting work, things like that. And then the second half of my career where I started getting into sort of the software tools and services were old. And that started when I joined Microsoft, around 2009 spent a couple of years working with Microsoft during the the HTML five and Windows eight era before jumping in to a company that made web tools, developer tools called TELRIC. And I worked for them for about four or five years. And that was really when I first started getting into product management started getting into more developer advocacy and team management. And one of the things that I've really loved over those, when I made that switch was getting into this place of being able to build tools for developers to actually really helped developers solve problems at their companies and for their businesses. And so I've continued that, about six years ago, I made the switch into the IoT world, first as a maker. And then as a professional, I think as many of us tend to do so started tinkering around with Arduino and Raspberry Pi and all those things as one does, and then had an opportunity to join particle, spend a couple of years working with particle on their developer platform developer community. And then just about two years ago, I joined a different IoT company called Blues Wireless, it was started by Ray Ozzie back in 2017. So done a lot of different things. But it is all really as it says, as I had written in my bio, it all really focuses on getting to help developers on uncovered unblock sort of new exciting things with technology.
Justin Grammens 3:48
That's awesome. And like I mentioned during the thing, I'm really excited to hear you talk about smarter everything, you know, using machine learning and the Internet of Things, because this is a space, you know, my background has been very similar, I guess, you know, got into the internet in the late 90s, you know, built a lot of web applications. And actually more into mobile, I did a lot of work on iOS and Android, but then kind of really embraced the internet of things over the past 10 years or so. But this new space really I would sort of call it the merging of machine learning and AI with the internet of things I think is really a fascinating space, sort of this concept around tiny ml, and I've seen some of your presentations that you've given at some of the NDC conferences and stuff like that. Are you working in that space?
Brandon Satrom 4:30
I am a lot and I have a real passion for AI and ML even though my company doesn't, we don't build anything that's directly associated with AI or ml. In particular, I have been a fan of the ML and especially the tiny ml space since I think I first saw Pete Warden talking about it on a stage at a Google conference, I think maybe four or five years ago now. And so I spent some time when I was at particle really getting into tiny ML getting into TensorFlow. I worked with a Google coral. I've done a bunch of work on doing sort of, I wouldn't even call it tiny ml on the coral or on the Raspberry Pi, but really starting to get into exploring AI and ML on edge devices. And one of the things that I love the most about working in this space, especially being part of blues is that we're we are focused on unblocking wireless connectivity, no matter what it may be, whether it's Wi Fi, or whether it's cellular, whether it's even technologies like Laura and Laura, when I do a lot of talks with our with friends of mine and Jim Paulsen, one of the things that we like to talk about a lot is that edge ml and cellular IoT go hand in hand, because we're moving away from this world where everything has to be streamed to the cloud, there's still a ton of great use cases for cloud based ml, especially around model training, model development retraining. But when it comes to actually doing inferencing, there's so many wonderful advantages to working at the edge. Not only do you have privacy, but it's also an opportunity to make decisions faster, even when cases where connectivity where there's high latency that we're connectivity is at a premium. And so I spend a lot of time and my team and I spent a lot of time working with using ml use cases as a way to show the value of bringing cellular IoT into solving many kinds of problems, whether it is you know, monitoring of existing analog devices, we've done some work on those kinds of things, whether it's adding mL at the edge to existing solutions, but you said yourself, Justin, one of the reasons why there is this merging of AI and ML and the IoT is because I remember my days at particle, we would talk to a lot of customers that would come to us and would say, I love what you guys are doing. I want to buy this, I want to use this platform of yours. But I also want to do some some AI ml stuff. Can you guys help me with that? And the answer to those customers is always well, if you're not already doing any data collection, there's not really anything that we can sell you right? Because in order to build to implement an AI or ml use case, you have to have something, you got to start with something and I want to give talks about this, what I always like to say is that the value, the raw materials in the AI, ml space are at the edge. They're in sensors, they're in devices, they're the things that we're looking to monitor. So step one, a lot of times in building out an AI or ml solution is being able to have that data, those raw materials in the form of what you're actually sensing what you're actually eating, how you're actually monitoring systems at the edge. And so I gave you a long answer that question, but it's obviously I love talking about this kind of stuff. And I think the technologies are really it makes a lot of sense that they're starting to become really hand in hand.
Justin Grammens 7:31
Yeah, no, that's great. I couldn't agree more what I've seen the Internet of Things happen over the past, I don't know, you know, eight to 10 years or so as I've been in this space is, you know, you have some companies that are embracing it. And they're they're willing to go all in and a lot of companies that that I have worked with iLab 651, have really sort of dip their toes into it right there. And they don't they don't see the value in the data, you know, quite yet. But once you start collecting that data, and you can make all these decisions at the edge, I feel like then companies will actually or have started to then adopt it. But do you feel like it's taken a while for us to actually be able to, you know, with the advent of TensorFlow light with the advent of better software tools, now we're starting to see the power of the Internet of Things where because I kind of feel like the IoT has been sort of just kind of stuttering here, spin honey has, you know, kind of like I wouldn't say in the next year, it's going to be the breakout year next year is going to be the breakout year. What's What, what's your thoughts sort of around that?
Brandon Satrom 8:26
Yeah, it's actually kind of funny, because I've done some writing on this recently in preparation for some talks that I'm giving later this year. But when I look back and analyze, I think if you looked at the Gartner Hype cycles, one of the examples that people use to show our technology trends are moving. The IoT first appeared on the Gartner Hype Cycle, I believe in 2012, or 2013. If you map its progress and regress on a timeline, it basically never moved past the peak of inflated expectations. And then ultimately, it was dropped by Gartner back in 2019. And so obviously, the hype cycle is not everything, but it is a good amalgam to tell this struggle we've had of understanding like, okay, are we going to have 500 billion devices? Are we going to get to a trillion devices? Some of projected Are we really just sort of waiting and like you said, it's seems like it's always on next year is the breakout year next year is the breakout year. And, you know, when we think about this, a lot of this comes back to complexity in the case of developers, right. Like, with the IoT, I think the biggest problem that has blocked people from being able to deploy is the complexity of getting a solution deployed. I usually refer to this as the strings of wireless IoT, the things that really will hold developers back that cut off choice and make it much harder for us to actually implement solutions. And it starts with things like having to learn at commands to work with cellular modems, and who wants to pull out a haze manual from 1986 to be able to actually talk to a network, you know, or, you know, having IoT solution providers that tend to constrain developers too much by forcing you into Their platform or their solution. And so what we have done at blues, wireless is really focused on bringing simplicity as much as possible to that story for developers. Because at the end of the day, what an IoT solution implementer wants is they want to take their sensors and their devices, and then as quickly as possible, get those things into their cloud, into their dashboards into their ultimate repository. And once they can do that, there's myriad things that they can do around edge ml, model, retraining, remote model updates, there's a brilliant set of technologies that are available at that point. But this this middle piece of secure, reliable connectivity, that is the problem and is continues to be I believe the thing that keeps us from not seeing that explosive growth that we have been projecting for years. And so our approach and the way that we believe the problem can best be solved this by taking almost sort of a socket modem approach with our products and the core product that we offer something called the note card, and it is literally just a cellular modem. And we have a Wi Fi equivalent as well. But the core product is a cellular modem that developers can communicate to and from, from their actual embedded projects using JSON commands. So no 80 commands, no certificate swapping, none of that, you know, typical exchange of information with a server. It's literally a device that already has a secure connection to our cloud service over 18 T's network. And it works on 135 countries. And all a developer needs to do is make a, you know, I squared C or serial connection to it, issue a couple of JSON commands, and you can stream data back into your device. And so I know I'm getting into the weeds a little bit. But I really think that creating simple clean abstractions on the top of what is a super complex problem when it comes to secure IoT connectivity is the thing that a lot of developers have been waiting for. So they can actually deploy pilots. So they can actually get sensors that are connected in the field. And I think once we have that point, once solutions can actually securely connect, collect their data, get that information in the cloud, it unlocks all of those edge ml use cases that I think we haven't really been able to capitalize on before because the data has continued to be trapped in these dumb devices that we haven't been able to instrument.
Justin Grammens 12:19
Yeah, for sure, for sure. Are you familiar with a company called nimble link? I am. Are you guys sort of must play in that same space with regards to essentially a modem interface? I guess, right?
Brandon Satrom 12:29
In a similar space? Yeah, yeah, absolutely. I mean, there's a ton of different providers in this space nimble link, obviously, particle where I was previously, there are some providers like Twilio, where you know the option of just buying the SIM and working with them on that regard. So in all of those cases, I think we really believe that we're developers, what developers really need right now is the simplest, no frills way of being able to actually use a device and not be tied in one of the other things I didn't mention that sort of core to the product as well is that when you purchase the no card, it actually comes with the data. So 500 Meg's of data over the course of 10 years, that's actually included in the cost of the device. And so there aren't monthly plans, that's actually another thing that I've seen and I have experienced as a developer is when I'm building a prototype or proof of concept, I really don't want to start the clock on a monthly fee, not just to validate my idea just to start collecting information. And so and I think that's actually one of the things that we're seeing in the space is that solution providers like blues and nimble Lake values are starting to really understand how they need to adapt their offerings to give developers a simple, no frills way of getting connectivity, because we want to make it possible for devices that are otherwise throw away or fungible, to be able to connect Pete Warden who I already had mentioned earlier, he uses this phrase, and he has he and I've talked, he likes to say Peel and Stick sensors, getting into this place where sensors and connectivity is so low cost. And so the way that I like to look at it is it almost sort of makes it possible to put an IoT device in a greeting card? Not that you would but at the same level of fungibility, right, you have something that is throwaway, it's single purpose, it may only live for a year or two. But the real important piece is the data that it's able to capture the thing that it's able to tell you about the world that makes everything else around it better.
Justin Grammens 14:15
When you talk about the peel and stick the first thing that my mind went to was RFID tags, right? So, so bad, you know, 20 years ago or so I remember being a part of this, an RFID is going to be everywhere. It's going to completely revolutionize everything. And that's another technology. I think that it found a different place. But it's not. I don't believe companies like Walmart, or Costco, all these places, were just going to basically RFID tag everything. And from my understanding, it never really got to that point. Right.
Brandon Satrom 14:42
I remember that as well. Yeah, in the early 2000s. I had a couple of friends that ended up, you know, starting RFID based companies and think they did fairly well, but it didn't have this sort of Cambrian explosion that I think you know, has been predicted was predicted for it. And I do worry sometimes about that being the same fate for the IoT. I mean, the prognostications versus what we've seen so far, hasn't manifested. But I still believe that the reason why is because of complexity on the part of the developer, right, that the promise of connectivity is there. But it has to be easy, and it has to be affordable. And I think once you can get to that point, then it becomes a no brainer to leverage the services.
Justin Grammens 15:18
So we've talked a lot about Internet of Things, maybe let's let's talk a little bit about artificial intelligence. One of the things that I like to ask people, I guess, is how do you define artificial intelligence? If somebody said, you know, give me a sentence or two on that? Or even say, what do you do in your day job? For example, you know, if you're touching on pieces of AI, or ml, how would you describe what that is,
Brandon Satrom 15:36
To me, it is about leveraging intelligence in the environment, leveraging the latest information that's in the environment around us to automate decision making, and to allow companies, customers, individuals to anticipate or solve problems before they start without manual intervention. A lot of my thinking about this is of course, bent towards the industrial IOT space, because that's what I touch, day in and day out. But I really do see that I think that's, you know, to me, a lot of the promise of AI is, when a business, for instance, let's just take some hypothetical industrial refrigeration company, the business of warranty management and service today, the state of the art is basically scheduling when trucks have to roll the service machines based on the average duration of failure for an individual device, right. And so you have some machines that will fail sooner, that's an early truck roll, you have some that will last longer. And the power of AI and I still believe this is true is always been that if you actually didn't base those timelines on an amalgam of all the data that you have about historical performance, but instead you understood or you, you trained a model to be able to build those correlations around actually, failures tend to happen often when this particular motor starts to get five degrees hotter. And when that happens, now it's time to actually be proactive about it. We call that predictive maintenance. Sure. And ultimately, I think that's a possibility. And there's a lot of power to that. To me, that's where AI is really interesting, because it helps us, it helps us understand correlations that are impossible, not impossible, much harder for humans to be able to pick out of all the massive data that we have.
Justin Grammens 17:16
Yeah, for sure. You know, there's this other term floating around Carl called the artificial intelligence of things aifd. And I think there are a couple of companies Bosch in particular, but I know they've started to sort of glom on to that out of that term, I actually just did a presentation last night, virtually to a bunch of MBA students, really around AI IoT, and sort of how it's different than just IoT. And the thing that I think about is is kind of reactionary versus proactive. Right. And so when you start when you start bringing machine learning and AI into the equation, you can be a little bit more proactive on these things, rather than being so reactive, when all of a sudden you've just sensed all this data, it's like, well, it's too late now. Right? Absolutely. You know, and so that's where I feel like maybe IoT is kind of in some ways, hasn't really fully matured to the standpoint of letting people see, hey, you can actually save a lot of time, money, effort, cost, whatever it be, you know, to be a little bit more proactive in your solutions?
Brandon Satrom 18:10
Absolutely. Yeah. You know, as with any form of thing that falls under that umbrella of what we like to refer to as digital transformation, I think a lot of companies, when they're looking at these things, will they tend to start by thinking in terms of new business, new revenue opportunities, new lines of business, and things like that. And that's useful, but I really think that there's a whole lot of opportunity under this idea of, of cost savings, and helping companies actually run their businesses more efficiently. It has a, an impact, a faster impact on the bottom line than it does on the top line, per se. And things like being proactive instead of reactive really helps with that. Because when you can talk in terms of truck rolls and service and warranty, you know, in redemptions and things like that, you can actually paint a through line to AI and IoT together, creating real savings for a company and helping them run more efficiently and deliver better returns to their own stakeholders and investors.
Justin Grammens 19:07
Yeah, for sure. You talking about industrial, you know, motors and stuff like that. Are there any other applications, I guess that maybe you've maybe you aren't even a part of maybe that you've seen in the news, I guess, where you sort of seen this overlap, I guess, of IoT and ml actually providing value.
Brandon Satrom 19:22
Another one that I think is really is powerful, and I did some proof of concept work on this last year is around leveraging machine vision in particular, but leveraging AI and ML to effectively add intelligence to analog systems without looping into the existence into the system physically. So this idea of sort of like the ultimate retrofit of being able to put a camera in front of a dial or a valve, whether it's an actual camera, I did some of that last year, the POC that I built was around the pool pump out in my backyard and being able to actually use AI and ML Walter Reed the gauge on the pressure tank so that I knew what I needed to backwash the filter, if there's an obstruction or anything along those lines, a lot of typical use cases will factor into looping into that system, taking out the analog sensor, adding in a digital sensor, et cetera, et cetera. But there's an interesting, I think this is still exploratory. But there's an interesting idea of being able to use leverage these technologies to just put a camera in front of something and not interfere with the existing systems. I have a co worker that's done some done some work around that with thermal energy, thermal imaging, and radiators as well. And so I think that's interesting. And the reason why the two tend to go together of courses are obviously the AI and ML concept of vision at the edge. But then there is the, the IoT piece of that is not only the backhaul, not only the ability to make a connection out to your ultimate cloud service. But I'm really fascinated with this idea of being able to leverage IoT connectivity to backhaul, raw data, even if you're performing edge inferencing to backhaul training results, perform retraining, and then perform over the air model updates back onto the edge device all using the connectivity that's there. So as your model is improved in the cloud, you can actually push those updates and that additional intelligence back down into the edge device so that it continues to get the benefit of fast inferencing. But it gets better, it gets better in a way that can't isn't possible today, unless you do training in the cloud in order to improve the model. So those are interesting use cases. To me, I love this kind of stuff. Yeah, for
Justin Grammens 21:34
sure. That is the internet in the internet of things, right. That's why these things are on the internet is so they can communicate with each other and get updated along the way. When you were talking about just looking at things I remember, we had a presenter come and he worked for a medical device company here locally, they were trying to essentially count the number of units were coming that were coming off the line. And they actually couldn't touch the equipment for for FDA rules and regulations, they actually couldn't update the equipment are really just because it's there's a whole there's just a whole bunch of quality management systems and stuff that are in place. So what they did was they literally set up a camera. And there was a motor that spun around each time one of these devices was kicked out and they put a.on it and they just watched the number of times the dot, the dot went around, right, it was sort of a unique way for them to be able to get the data that they wanted, be as you know, unobstructed as possible and still sort of get what they want it done in a sort of unique creative way using computer vision. So I know what you mean about that.
Brandon Satrom 22:33
It's funny because I saw Stacey Higginbotham had written an article last week about I think she said tiny ML is still looking for its killer use cases. Right. And I think that I understand that headline, I think the article is pretty good. She wasn't saying that there is no application to this process. But it almost kind of makes me wonder if you know, as you ask the questions, Justin about AI IoT, that as we're looking at the IoT, still sometimes struggling to sort of cross into the, into the slope of enlightenment, again, back to Gartner terminology. And if tiny ML is still in that place of that sort of that storming phase of understanding where it's applicable, a Iot might actually be the answer here that it's really the combination of these technologies that the industry has been looking for, because it's the, it's the connectivity, that's essential. But it's the application of that connectivity at the edge, to be able to anticipate problems just to provide insights and information to customers and businesses that are looking for it. I think that's a thought that I hadn't had until until we just started talking. But I think there's an interesting marriage that might be the solution to these technologies really, really getting into into a positive place for large companies.
Justin Grammens 23:41
That's funny, you mentioned Stacey, I'll have to listen to that. Because she was actually at the tiny ml summit that I was at a couple of weeks ago. And I actually sat at the table with her when we had breakfast. And it was interesting, I didn't sense negative vibe at all, really, we were just sort of talking about the sessions and a lot of the sessions there, I could sense that they were still probably three to five years out from being commercialized. People were talking a lot about different ways that you could train models, and have them applied with very, very low power, right, it was it was all about power consumption. And these are PhDs. These are people that are part of research organizations that are funded by X, Y, and Z. There's some really cool platforms you mentioned, you know, Edge Impulse, they're pretty much a leader in this space. They were they were pretty much the big dog there. But there are a lot of other companies that are are getting into this space as well. But it was very interesting to see the difference between applications that are people that are people actually are, I would say platforms and service providers, and then research and the research guys were very much focused on tuning, optimizing all sorts of white papers and all this sort of stuff about it, but it hadn't really it hasn't, in my opinion moved its way into what the actual industry is doing today. And so it was fun to see it was a great, great conference, I would absolutely go back again and I will definitely have to take a look at what she was saying. Hang, but I think I sided with her a little bit on that. Because, you know, with any new technology, I think, is it still trying to be figured out? And where where is the, I guess, where's the killer app? You know, for some of these things?
Brandon Satrom 25:11
Absolutely. And we talk about that a lot. I mean, yeah, I don't think that the you know, I think the timbre of the article was not negative in the sense that tiny ML is not going to get to that point. But that's just sort of an assessment of where you know where the space is, in terms of where we all believe that it that it can be. And it's not surprising that that's really still where a lot of the research is, because if you think about it, I mean, the genesis of tiny ml was wake word detection on mobile devices and smart speakers. And so I think, in many cases, understand that is a very key use case, and is a key killer app in many ways, but are looking for what's the you know, what's the industrial killer app equivalent? What's the, you know, the mass adoption killer app equivalent, I think a low power research is going to benefit. This is actually an interesting thing for us at blues, because low power is part of really everything we do. We you know, as odd as it sounds, what we built with the no card is designed to be in a low power device that runs on cellular, which was really unheard of as recently as a couple of years ago. So this idea of being able to not only have edge devices that can run low power that can run almost in sort of a DSP user space, but at the same time, have cellular connectivity, that's able to use the power it has to to connect to the cell network, because that's somewhat constrained by the laws of physics, but at the same time, can sleep for long periods of time can allow a device to run off a battery and solar power really, as long as it needs to. Because again, if we're going to build use cases with this idea of allowing companies not to have to roll trucks on a schedule, and yet we build devices that need batteries that have to be changed every six months, we haven't really solved, we've traded one problem for another one. And so I think a lot of that research will continue to benefit but I agree there's still this thing, there's still this latent desire to find out like, what are the half dozen things that are really going to define this space in this industry? For us?
Justin Grammens 27:07
Yeah. Well, one of the things that I also like to talk to people about is is, you know, how would this technology affect future work? Or how was work being done by humans in the future? potentially less truck rolls means there's less truck drivers, right? Do you view some of these things having a negative impact, I guess, on the workforce of the future?
Brandon Satrom 27:26
That's such a tough question. It's a transformation in terms of how the kinds of jobs that people do, I would never argue that there won't be an individual negative impact, right? Because that's always a possibility. But if we anticipate these things, if we're open and upfront about them, it's not about AI, putting people out of work, as much as AI enabling people to do work that matters more, that's more important. It's more critical. So I don't buy into this idea. I'm very skeptical of the idea of a generalized artificial intelligence at any point in the near future, if if at all right? We don't have a complete understanding of our own brains. I don't know how we would create intelligence using that limited understanding. But I do think that when it comes to job displacement to roll this placement company should approach that with an opportunity to find out okay, what are the things that we can't do today that this enables us to do? What are the things whether it's in a whether it's in r&d, whether it's through lines of business, whether it's just changing the way that we think about, think about work, there is always an opportunity to leverage technology in a way that frees us up to do things that we that were either impossible or really difficult to do before?
Justin Grammens 28:33
Yeah, I totally agree, totally agree with that. What is a day in the life of a person in your role. And then also, as a follow up, you know, people looking to get into the field, what's in any sort of advice that you would suggest people take with regards to books or conferences or other things.
Brandon Satrom 28:50
So typically, for me, I have a pretty interesting role. And it's actually I love what I do, I get to sort of do a mix of leading a team and also getting to do a bunch of hands on work myself, that's something that I've learned over 22 years is kind of essential for me. And there's a couple of different engineering teams that I work with pretty closely day in and day out. And we're typically working on hardware, firmware and web applications that are designed to help accelerate our customers and their adoption of blues technology. So whether it's a IML use cases or or in a more general IoT connectivity use cases, we try to really make sure that customers can do more than just buy a dev kit from us and not really know where to go. And so we tend to do a lot of work in that space. The developer relations team is part of my organization as well. And so I spend a lot of time working with our developer community working with our our documentation, our getting started experience, really across the board. So a typical day for me, will usually involve some form of working towards an event or workshop or something else or just speaking engagement that's up and in the offing. For me the current one right now is that the end of this month, I'm myself and the head of DEV Well, a colleague of mine are going to be traveling out to Portugal for the NDC Porto conference. And we're doing a two day workshop called, basically machine learning from the edge to IoT. And we're walking or over the course that two days, what we're going to be doing is walking participants through an introduction to ml, AI, AI and ML, for those that have never worked on it before an introduction IoT, and then how we can actually bring the two together, we'll be doing some work with TensorFlow light, we'll be doing some work with Edge impulse, and really just introducing as many developers as we can to the opportunities in the space. And so we actually are going to be doing that workshop four times this year over the course of four different events. And so there's obviously a ton of work still to come with that, that workshops at about three weeks, and we're you know, as usual, we're behind the eight ball, because that always happens. But we'll get there. So there's a lot of work right now to sort of get ready for those kinds of things. And that's the kind of stuff I love is just walking, giving developers an opportunity to, to understand and learn those, those new technologies. There's a few other things on my plate, you know, we do a lot of work on my team, a lot of the stuff that we do, as examples and samples for developers, we publish on hackster.io. That's a channel that we absolutely love. And so we typically will do a couple of extra projects per month, a lot of times they end up being, not all the time, but I mean, about half the time there ends up being an AI or ml component to them, because it's just an area that we love to work. And we see a lot of interest from customers. So it's fun to get to do those things.
Justin Grammens 31:26
That's awesome. So yeah, I spoke at NDC when it was here in Minneapolis, probably man four years ago, or something like that. It was quite some time ago, you know, Microsoft had actually put out this dev board by MX Chip, I think and it basically connected to their as your back end. So I did a whole session with and some some very simple listening for sounds, sound detection, movement, stuff like that. So you know, sort of showing data that you could quickly get to the cloud. That's awesome that you're going to Portugal, I had actually submitted a couple of talks, because they happen all over the world. Are you going to any other cities as a part of the NDC?
Brandon Satrom 32:00
So we're doing Portugal in April, Copenhagen in May, and then Oslo in September, which I've spoken in NDC, Oslo three times previously. And that's the original NDC conference. And I love that one. So yeah, I'll be doing that one. And then there's a few other events in the US that conference is a developer event that typically takes place in the Wisconsin Dells every summer. So that's going to be in July, I'll be up doing a workshop and talk at that event as well.
Justin Grammens 32:27
Are you you're you're going to be at that is going to be a conference? Yeah, yeah. How are you? Oh, awesome. Yeah, no, I sent in a kind of a hands on thing with IoT, as well, to Clark. And then last month when they had to sort of call for speakers so Oh, nice. Oh, yeah. Sorry. But you're going to be kind of doing this this this ml IoT thing?
Brandon Satrom 32:46
We're gonna do a half day workshop. I yeah, I don't know if it's gonna be an ML IoT. We haven't decided on the topic for that one yet. Because it's still kind of a little bit ahead. I think we're gonna adapt our two day workshop into more of a half day kind of thing. So yeah,
Justin Grammens 33:00
very cool. Yeah. No, I've not been to that before. You know, let last year was the first time I think they brought it back to Wisconsin Dells since the pandemic, and then and then this year, yeah, I submitted a talk and I haven't heard anything back. So I'm assuming it. It wasn't accepted. But I'm going to try and go. Barring all the other stuff that my kids do in the summertime. That's, that's what's hard, is they already signed up for day camps and stuff like that. But Oh, that's awesome.
Brandon Satrom 33:23
That is always tough. Yeah. Yeah. Well, I don't know when they're gonna announce all the speakers and everything. So there may still be time. But I have been familiar with that event. For a long time. Clark is actually a friend of mine. We used to work together at Microsoft way back when so we've got over the years and
Justin Grammens 33:36
that's cool. Yeah, yeah. No, I just I just kind of found out about that community. Just only a couple years ago. I was like, How did I not know about this? I'm kind of in my own Twin Cities bubble. Right. And you're you're Where are you over Madison?
Brandon Satrom 33:47
I'm actually in Austin, Texas. That's where I'm okay. Yeah. Okay. But so yeah, for those that are listening to podcasts, y'all check out that.us. Because, yeah, it's an event. It's been around for a while, but they have been really growing in the last couple of years. And it's, it's a good set of events,
Justin Grammens 34:03
For sure. And they have one in Austin right there.
Brandon Satrom 34:06
There's one in Austin in May, it was gonna be in January, still a challenging time. So we they moved it back to me, and I'll be speaking of that one as well not doing a workshop. But I'll be doing it. I also do a series of talks on using Python on microcontrollers, because of course, I do a IML stuff. Of course, I love Python, and I love circuit Python and micro Python as well. So I do some talks on, you know, showing people how they can leverage their Python language skills on microcontrollers as well.
Justin Grammens 34:33
That's awesome. Well, yeah, so I do include links and stuff in the liner notes for this for these podcasts. So I'll be sure to get both, you know, links off to you and any other stuff that you want to share with me at the end of this so people can take a look at it in the notes. This is so great that you're getting out in the community and sort of sharing showing examples doing workshops. You know, I'm not sure if I shared this with you, but I I teach a class on IoT at the University of St. Thomas here in St. Paul. I've been doing that for number of years when the pandemic hit, it's very hard to teach people this was this is graduate students that are going to a software like a master's program in software to, you know, to use breadboards and plug in things like that. So kind of talk a little bit of a hiatus over the past year, but I'm coming back in the fall. And I'm not going to do the same old IoT class. Now, I'm actually kind of going to be using the tiny ml book by O'Reilly, the that Oh, that's great. That guy, you mentioned that, that's kind of going to be my syllabus, in some ways. It's like, you know, we're gonna learn. Yeah, so the class is like 16 weeks long. And we're going to use the first part kind of building some stuff on particle just to kind of get the data to the to the cloud. But now let's actually work on actually running in a machine learning at the edge. So I'd love to, you know, I think we should talk offline a little bit more about sort of how you're structuring some stuff. I'd love to share with you sort of what I'm thinking about as well.
Brandon Satrom 35:47
But oh, yeah, I'd love to. Yeah, that's great. And yeah, that the tiny ml book is a fascinate is a phenomenal one I have I got a copy. I think I got like an early copy of that. When Pete and Daniel first started writing it. And that's actually one of my favorites. I think you had asked a little while ago about books and whatnot. I didn't want to forget, probably my absolute all time favorite is the grokking deep learning by Andrew Trask. That is, it's a maddening book. And it is I absolutely love it, he actually walks from first principles of understanding deep learning outside of any frameworks all together. So you basically build your own sort of first simple deep learning models and convolutional neural networks using just Python and NumPy. And then you sort of build up from there. But it's a really good sort of hands on kind of book. That was one of the first ones I really dug into.
Justin Grammens 36:37
Nice. Yeah, that's great. I will definitely look that up and include that in the notes. How can people connect with you find you on LinkedIn, I guess Twitter?
Brandon Satrom 36:45
Yeah, LinkedIn, Twitter, I'm Brandon sat be satrom on LinkedIn, and Brandon's Hadramawt, on Twitter. And both of those places, I have a blog that I keep meaning to actually update. But yeah, you know, yeah, other stuff to do. But I do. I do write pretty regularly on the on the blues blog, if you go to blues.io. Our blog there has a fairly good set of information. And I publish projects on hackster.io as well, where I'm also there as Brandon Satrom.
Justin Grammens 37:12
Very good. Very good. Yeah. Is there anything else that you wanted to talk about? Maybe that I didn't touch on?
Brandon Satrom 37:17
No, I think this is this has been a great chat. I love I love talking about AI ml, I'm glad you brought up the the AI IoT concept. I know it's kind of get its buzzword treatment. But I think there's a lot of validity to the to the marriage of these technologies, and really taking us to new heights.
Justin Grammens 37:33
Awesome, for sure. Well, yeah, I'm looking forward to you speaking on July 7, at our applied AI meetup. And that will be at 630 Central time, for those that are listening. And I'll be sure to get all the word out to everybody here as we start to planning for for that event. But yeah, Brandon, thank you so much for the time, I appreciate it. And yeah, I look forward to keeping in touch with you and sharing more and more about, you know what this what's going on in this space. There's a lot of fascinating things. It's always evolving, always changing. So that's what I really enjoy. And I really enjoy having guests on the program and just sort of talk through with them what they're seeing, so I always learn something.
Brandon Satrom 38:10
Yeah, thanks, Justin. It was a great conversation.
AI Announcer 38:14
You've listened to another episode of the conversations on applied AI podcast. We hope you are eager to learn more about applying artificial intelligence and deep learning within your organization. You can visit us at applied ai.mn To keep up to date on our events and connect with our amazing community. Please don't hesitate to reach out to Justin at applied ai.mn If you are interested in participating in a future episode. Thank you for listening