In this episode of the IoT For All Podcast, TensorIoT’s Software Manager and IoT Practice Manager, Shahan Krakirian and Joseph Melendez, join Ryan Chacon on the podcast to discuss how to pick the right tech for your IoT solution. The podcast opens up with Shahan and Joseph talking about TensorIoT and the use cases of their work before getting into how and when TensorIoT joins a company’s IoT journey. They then discuss picking the right tech for your solution and how it can fit into a business without overhauling company processes. Ryan, Shahan, and Joseph wrap up the podcast with a conversation about the importance of the cloud in IoT applications and when edge computing or hybrid solutions may be more applicable.
Shahan Krakirian helps the Product Team to develop and refine TensorIoT products. His favorite part of the job: “Many jobs are repetitive – mine is not. We’re constantly challenged (and growing) in an ever-changing environment, but always given the support to excel.” Outside of work, Shahan enjoys rock climbing.
Joseph Melendez is a software engineer leading the IoT team. He’s dedicated to addressing customer needs, working in multiple programming languages, having expertise in Amazon Web Services, and hands-on experience with the hardware needed to bring their vision to life. Driven to learn from new experiences continuously and eager to address questions and concerns in the ever-changing tech landscape. He graduated from the California State Polytechnic University, Pomona, with a major in Computer Engineering and a minor in Mathematics.
TensorIoT was founded by Amazon and Google engineers. They focus on IoT, AI, and serverless solutions. Located in California, their team of architects and engineers have extensive experience in rapidly developing high-velocity IoT/Edge solutions powered by AWS services, including Sagemaker and Alexa for Business. Their AWS Advanced Consulting Partner status showcases our ability to harness Amazon’s powerful ecosystem to increase the scalability, availability, security, and performance of solutions. Leveraging our deep AI/ML expertise, we have built open-source cookbooks which bake in best practices and allow our team to rapidly deploy business impacting solutions for our customers.
Key Questions and Topics from this Episode:
(01:33) Introduction to Shahan and Joseph
(03:16) Introduction to TensorIoT
(04:36) Use cases of TensorIoT
(07:18) Typical client interaction
(10:04) Picking the right technology for your solution
(13:20) How IoT fits into a companies current process
(16:14) Integrating the cloud into IoT
– [Voice Over] You are listening to the IoT For All Media Network.
– [Ryan] Hello everyone and welcome to another episode of the IoT For All Podcast. I’m your host Ryan Chacon. And on today’s episode, we have two guests from TensorIoT, Shahan Krakirian a Software Engineer and Joseph Melendez and IoT Practice Manager. For those of you who may be unfamiliar with TensorIoT, they were founded by AWS and Google engineers with a focus on IoT, AI and serverless solutions. Their team has extensive knowledge and experience rapidly developing high velocity IoT and edge solutions, all powered by AWS. Great conversation today. A lot of fun talking to both these individuals. We talk a lot about the process, strategy and advice for companies picking the right technology for a particular use case. We’ll walk you through the process, what companies should be looking for, how you can prepare, what are our best practices, general advice, that kind of thing. We also talk about current cloud versus on-prem options out there in the market. Why companies should be adopting cloud, what the advantages of doing so and so forth as it all relates to IoT. So all in all great conversation, I think we’ll get a lot of value out of it. But before we get into this episode, I want you to take what you know about IoT and expand it by a factor of a thousand. That’s what our sponsor for this episode Impinj is doing. With the Impinj platform, everything can be connected. From strawberries to supply chains, from IV bags to inventory, Impinj is creating the internet of every little thing. To learn more, go to impinj.com/ifa. That’s Impinj with a J, impinj.com/ifa to learn more. And without further ado, please enjoy this episode of the IoT For All Podcast. Welcome both of you to the IoT For All Podcast. Thanks for being here this week.
– [Shahan] We appreciate you having us, Ryan.
– [Joseph] Thanks, Ryan.
– [Ryan] Yeah, I’m very excited about having this conversation. I always love it when I can get two guests on. Get two different points of view and a good chat going. So I wanted to kick this off by having you both introduce yourself to our audience. Just a background experience. Anything you think could be relevant for our audience to get a better sense of who we are listening to.
– [Joseph] Yeah, that’s good.
– [Shahan] Okay, great. I’ll go ahead and start. So my name is Shahan Krakirian. And I’m a Software Engineer with TensorIoT at our company. I’ve been there for about three and a half years now. And my primary focus over the last, especially a year and a half or so as Industry 4.0, so Industrial IoT work has come around. That’s kind of been my focus mainly, yeah.
– [Ryan] Fantastic.
– [Joseph] And as for myself, I am the IoT Practice Manager for the company. I’ve been with the company for about three and a half years, almost four years at this point. Joined at the time when the company was still very young and starting to grow. Since then we’ve definitely had a chance to dive into multiple really awesome projects to explore some cutting edge technology. We’ve been able to expand our team including hiring overseas. We got a Japan team, India team, UK team. And Shahan and I have both been part of that journey and seen the company grow so-
– [Ryan] How big is the company now? How many people?
– [Joseph] I think we’re about 100-
– [Shahan] About 100. Yeah, 100 or one 110, something like that. I think Joseph you were employee what? Eight or something?
– [Joseph] Yeah.
– [Ryan] Very cool, very cool. All right, so TensorIoT. Let’s talk about it for a second. Joseph, would you mind kind of giving a quick introduction to what TensorIoT is all about, the role you all play in IoT and feel free to go from there.
– [Joseph] Absolutely, so I mean our company, just to kind of give a high level overview. We designed the best in class products and solutions to connect devices and make them intelligent. Our slogan is making things intelligent. We really believe in the transformative, potential smarter devices. And we really think that that could enhance our quality of life. We were founded by a former AWS employee. And all our employees are certified. We’ve got numerous competencies. Definitely a larger amount of our focus is on IoT and machine learning workloads on the AWS platform.
– [Ryan] So when you’re talking about connecting kind of devices, talk a little bit more in detail about kind of the areas in which you focus on. Are there also any vertical specific areas that you are more in line with or use cases that you aim to work on over others?
– [Joseph] Definitely, I think we kind of cover a little bit of everything as a consulting company. It really depends on the client’s needs. We’ve worked agriculture, oil, mining industry. We’ve done a little bit of everything. We focus on different technologies.
– [Ryan] Okay, fantastic. So I guess breaking that down a little bit further when we’re talking about the use cases. Are there any that either of you comfortable kind of talking a little bit more in detail about? And what I’m kind of getting at here is just to kind of bring it all full circle to showcase the role you all play in those use cases and some successful deployments that you don’t mind sharing.
– [Shahan] Yeah, sure. So we’ve got a couple of public references that we can share.
– [Ryan] Perfect.
– – [Shahan] So one of them is a little company called Thermo Fisher. So they make lab equipment and medical devices. An example of something that they wanted to accomplish was they’ve got refrigerators and freezers that are obviously very purpose or purpose built for particular experiments. So if you can imagine, if you need to have some sort of solvent or something like that in a refrigerator at a particular temperature, then that it needs to be that way. And it would be better for the organization if you didn’t have to have somebody standing there, making sure that your refrigerator is at a particular temperature for X amount of time and waste that resource where somebody could be going and doing something else. So the thing about these machines, but that they were fairly old. So our goal was to go in and kind of modernize and bring the data from these refrigerators and freezers into the AWS Cloud for kind of a centralized viewing space. So you can see them in a user interface and also so that you can be notified if something is not going the way that you want it to go. So that you can kind of dispatch people to get there quickly and solve the problem. So TensorIoT was integral in this by deploying hardware and actually connecting in via serial port. But the hardware that we had deployed on site, had a bunch of AWS Software on it that enables connection to the cloud, send in some telemetry data capturing not only information about the device itself, the hardware, but also about the refrigerators and the temperature and the pressure inside and that sort of thing. So we accomplished that goal for them. They were able to not be standing around, standing around looking at their experiments. They were able to use this device called the Devicelink Hub which is what we built for them.
– [Ryan] Oh, fantastic. So take me through, what does a typical engagement look like for a customer outta curiosity? So when they come to you, is there a certain stage of development or stage of like on their IoT journey they’re usually in and what does that kind of relationship usually look like?
– [Joseph] Sure, typically when a client comes to us, a couple of things, we’ll ask a couple questions just to gauge where they’re at in their journey. Whether, it’s a brand new idea that they’re trying to start from the bottom up, or whether there’s something in place already in existing infrastructure that they’re looking to build off of. And typically, we kinda have a process. We have a couple questions we ask. We really try to extend a discovery phase to make sure that we’ve really honed in on that scope. So that we can define what the client’s trying to do basically.
– [Shahan] And I think over the last couple of years when we, or at least this is my experience when I started off at TensorIoT, many of the customers that we were working with were kind of very, very early on in their IoT journey. So we have almost no visibility into our infrastructure. Can you get some hardware, put it in there for us? But slowly but surely, at least in my experience, I’m seeing that some of the customers actually have a bit of an infrastructure. Maybe they’re already collecting data. And what they’re looking to do is to enhance or perform better analysis on that data in some cases.
– [Joseph] That’s even though- Oh, sorry, go ahead. I’ll let you-
– [Shahan] I’m good.
– [Joseph] I was gonna say, one use case that I think really highlights that early phase of our company is, our work with Reliance Steel and Aluminum. Because at the time, they had these large machines that had been in service for decades. There’s really old machines that there was no real interface to tie into them and get information from them. So what we did with them is we took our own sensors and installed them into this various machinery. Because at the time it was hard for them to determine what machines was gonna fail apart from after the fact. Once the machines broken, it’s broken. And at the time, you basically end up losing productivity at that point and there’s huge loss of revenue from having that machine go down. So what we did is we put these sensors inside to measure different things. Temperature, humidity, vibration. Even some cameras in there. And we were able to create machine learning models off of that information that we collected over some time period. To do that but, coming in, there was no real hardware in place that could get that information.
– [Ryan] Right, right right. Well, that’s fantastic. So I wanted to actually kind of, it doesn’t necessarily tie directly into this use case, but I think from a high level when we’re talking about the engagement and the work you do with companies and guiding them kind of through the entire IoT journey and processes that they go through, I think one area is talking about, it is also oftentimes kind difficult for a lot of these companies it’s choosing the right technology. Cause it’s not just one type of technology they have to use for any particular use case. There’s connectivity, there’s the hardware components, there’s the software, there’s the cloud versus on-prem. There’s tons of different areas where technology comes into play. And for most companies that seems to be a step in the process that is very difficult for them to do themselves. Evaluate correctly, price it out correctly, understand what the right fits are for a particular use case. So I’d love it if you could just take us through what that kind of process, strategy, what advice you have for companies that are kind of in that phase of working to pick the right technology for a particular use case. What should they be thinking about, how to approach it, how to prepare for those discussions, things like that.
– [Joseph] Absolutely. I think the first step is kind of figuring out the overall goals and requirements for the project. And I think that’s one of the most important steps because there’s difference between asking someone what their use cases and they say, I wanna measure temperature. That doesn’t really help us hone in on a specific solution. Kind of figuring out those specific requirements. Maybe they’re trying to measure things every 10 seconds. Maybe it’s every, a minute, every hour, maybe it’s cold environment versus hot environment. I think getting those goals and the specific requirements is one of the big first steps. And with that comes something we mentioned earlier, is there an existing infrastructure or is this an entirely new implementation? I think, go ahead, sorry.
– [Ryan] Oh no, I’m good. I just didn’t know if- I guess, how can people best prepare for that when they may not be familiar with lots of the technology? Is that something that there’s resources that you kind of push them towards? Is that something that you all just more or less help them scope out? What are your thoughts there for people listening?
– [Shahan] Yeah, I mean that typically the first depends on the scope of the project, the size of the project. But that’s what we do for the first week or two in a lot of cases. We’ll sit down and get together with a company and kind of have them- Or we methodically go through, okay, what sort of industry are you in? Are your devices going to be very far apart? Are they going to be very close together? What sort of protocols can we use to collect the data? What does the data look like? Is it relational? Is it non relational? Can it be represented in a graph kind of structure? So I guess the answer to your question is sometimes people have folks in house that are somewhat knowledgeable but what we typically bring in is the integration of IoT with the AWS platform and that’s kind of our specialty. So that’s where we come in and help how to integrate into the cloud and build solutions on the AWS Cloud.
– [Ryan] And a lot of organizations obviously have existing infrastructure, existing technologies in place. Is there kind of an approach you recommend them taking or a way to kind of think about the legacy systems that are in place and how that’s usually handled in the discussion? Cause I think some people view that as a hesitation on how it’s gonna interrupt their existing infrastructure, existing technologies, they have deployed. But from my experience, talking with a lot of experts on the podcast, it sounds like it’s a very solvable solution and something that people should not be necessarily hesitant or scared about, but more so see it as an opportunity to improve and enhance what they already have, and working with an organization like yours to really navigate that in a successful way.
– [Shahan] Yeah, I think, working with working with us, one of the benefits that we bring is we’ve seen a lot of different types of implementations. And across companies that have legacy equipment. We actually work with another partner, CloudRail who essentially what their responsibility is, is they have essentially a plugin solution where you’ve got a number of sensors that record a number of different metrics. So it can be vibration, it can be temperature, et cetera. But it’s a plugin one click solution essentially and it integrates natively with the cloud. So that’s one option, but of course you do have to have those discussions where okay, if I do my installation or if if we’re going to install something ourselves, do we have to do this during off hours? When is your factory running? When is your industry down? That sort of thing. It really, really varies. IoT is an extremely interesting field because when people ask you to describe it, it’s really difficult to do because it applies to almost everything
– Ryan] It does.
– [Shahan] It’s more like a tool than it is a thing in itself, so.
– [Ryan] Agreed. I think the best way I’ve found to describe it when talking to individuals who work in enterprises and commercial companies and industries is trying to relate it back to hen application that could connect to them and what they do. You can obviously break it down at a very fundamental level, but it doesn’t always resonate. And as you kind of extrapolate that out even further, you start to really see the value that, and oftentimes a lot of these companies already have IoT involved in what they’re doing, they just don’t realize it. But when you kind of explain the value in what they could be doing that they’re not already doing, it’s amazing to see the surprise peoples have on the capabilities of what IoT can bring to them.
– [Joseph] Yeah, absolutely.
– [Shahan] Absolutely.
– [Ryan] So I did have another question I wanted to ask you. So you mentioned the company was founded by AWS employees, and so you’re heavy on AWS and the tools there. When we’re talking about cloud and integrating the cloud into an IoT solution, how do those discussions usually go? And by that, I mean, are most people very much well versed and knowledgeable about the value working with the cloud has versus being kind of an on-prem option? And if you wouldn’t mind kind of breaking down both of those sides as to the advantage and disadvantage of going either route and why companies should really be adapting to the cloud side of things and the advantages of doing so.
– [Joseph] Sure, I think oftentimes discussions depends on the people that we’re talking to of course, and their familiarity with it, their comfort with it. Oftentimes there’s a couple of reasons people are really hesitant about going off-prem. Sometimes it’s just a requirement for the industry to require data to be stored on-premise. Maybe they don’t trust the cloud quite yet and they don’t quite understand how security works in the cloud and the different things you can do to secure your data. When you hear about some of these things that happen, it’s usually not because- The security leaks don’t always tend to happen because the cloud didn’t have the capability to secure their data. It’s often cause of a mis-configuration. So having someone guide them through that and help them get up to speed with the offerings, I think is very important. But another reason some people choose to kind of stay on-prem is latency issues, of course. When you think about network connectivity, some operations need to be running live. You can’t have time for that that latency factor that comes with sending information out to the internet. But with those two points that I mentioned, I think there are viable solutions for this. In particular for having to keep your data on site. There are very various hybrid type architectures that could be used as a solution for this. For the latency issue I described, a lot of these different cloud services offer an edge compute capability. For example, with AWS, there’s something called Greengrass that works as a store and forward mechanism. So you don’t have to worry about if your internet goes down, things are still operational.
– [Ryan] Right, right. And have you seen any particular, I guess, verticals that maybe- And a lot going towards the cloud obviously, but do you see any particular verticals that are kind of trying to stick more with on-prem or maybe hybrid and hesitant to move more cloud based? Or are you kind of seeing a pretty good shift across all verticals towards cloud when it makes the most sense in that case?
– [Shahan] Yeah, I think generally this shift, I think we’ve passed that hump where people. I think most people are fairly comfortable slowly starting to move over. I think the question is, how much are they willing to invest in that right now? So for example, we’ve worked in the gaming industry, so casinos. That industry seems to be a very, very slow moving and kind of older school type of type of environment whereas, others you do have to be very careful, but they are willing to move over. Like medical devices, for example. There’s just so much potential there. Obviously, you have to be very careful like Joseph mentioned and take all the security precautions and things but it varies from industry to industry. But I think generally people understand that this is the direction we’re gonna go sure in the future.
– [Ryan] And are there any use cases that you actually would probably say fit better for on-prem versus either hybrid or cloud based? I know you kinda gave the parameters for kind of how to think about it, but I just didn’t know any particular use cases that you would say, Hey, this is one that like, it’s very hard to go away from the on-prem model.
– [Joseph] I would say one thing that we see sometimes is at systems that have what do we call, PLCs on site. SCADA systems, data visualization, that live visualization. There’s a bit of hesitation there sometimes but given some of the solutions and some of the more recent tools, I feel like a lot of these systems, people are figuring out ways to get that data over to the cloud while still keeping it on premise.
– [Ryan] Okay, fantastic.
– [Shahan] Yeah, that’s a great point actually. Sorry, Ryan for interrupting.
– [Ryan] No, no, no. Please, please go.
– [Shahan] Yeah, those systems are, they’ve just been built, there’s so many different software vendors that you can use to represent that type of data and they all represent it so differently. So kind of consolidating that is an extreme challenge. Because in one operation, you might have three different SCADA systems for example, that- And SCADA is kind of like the way you collect, the way you store, the way you visualize all of that stuff for manufacturing and industrial. So to make a digital representation, that’s kind of unified across all of those different SCADA systems and the way that they represent data is an extreme challenge. And that’s kind of the whole industry 4.0 smart factory moves right now. Which is what I’ve been working on recently. So that’s a big challenge.
– [Ryan] Fantastic, yeah, that makes a lot of sense.
– [Shahan] Solve that one, whoever can solve that one in a repeatable way, great stuff.
– [Ryan] Yeah, congrats with you. Yeah, fantastic. Well, this conversation has been a good one. I wanted to ask on the TensorIoT side of things, anything that we should be able to look out for coming out of you all in the coming months. Anything maybe in the industry in general that you’re looking forward to seeing. Just stuff that might be exciting for audience to be able look out for.
– [Joseph] Absolutely. I think one of the big things that happened recently is we hired a director of hardware solutions, Myron Hattig. And so we’re really starting to expand our hardware expertise and jumping more into creation of custom hardware. We have our offerings, our smart building kits and asset tracking kits. But we’re continuing to build up on that area and I think that would be great.
– [Ryan] Fantastic, and for audience out there, who’s listening wants to ask follow up questions, learn more, check out kind of everything you have going on. What’s the best way to do that?
– [Shahan] We’ve got our website, www.tensorIoT.com. You can go and look at the, get to know the team members a little bit. You can see what we do in some of our offerings. So that’s one place you can contact us there. Otherwise normal social media outlets. We’ve got LinkedIn, we’ve got Instagram and Twitter. And LinkedIn is probably the best place if you wanna reach out and talk to us in more detail about what we do. We’re a very broad, apply our skills to very broad set of industries and things. So our engineers are happy to share with those experiences. So go ahead and reach out to us on LinkedIn if you want.
– [Ryan] Sounds like you all love the challenge regardless of what questions, stage of development they may be in, use case or ideas that they may have. That’s awesome.
– [Shahan] Never the same.
– [Ryan] That’s the beauty of what we do. But thank you both so much for taking the time today. This has been a great conversation. I liked how we were able to break down kind of that process for choosing the right technology, learning a little bit more about what you all have going on as well as the cloud versus on-prem kind of thought process and models there. So thank you both so much for taking the time and we’re gonna be doing some more content that I think is gonna be right up your alley that I’d love to have you back for to, to kind of dive more into some details around some of these topics we talked about today.
– [Joseph] All right, awesome. Thank you, Ryan.
– [Shahan] Very cool. Yeah, thanks for having us, Ryan.
– [Ryan] Thank you. All right everyone, thanks again for watching that episode of the IoT For All Podcast. If you enjoyed the episode, please click the thumbs up button, subscribe to our channel and be sure to hit the bell notifications so you get the latest episodes as soon as it become available. Other than that, thanks again for watching and we’ll see you next time.