Screaming in the Cloud - How to Responsibly Automate Your Home with Mike Gray

Episode Date: September 24, 2024

From elementary school music teacher to a Senior Cloud Engineer at Defiance Digital, Mike Gray has lived quite a few lives. He hit it off with Corey during the AWS New York Summit this past summer. What brought them together? Their mutual frustration at what dominated the discourse of the event: the current fascination with GenAI. Although Mike has his qualms with AI, he also enjoys working with it quite a bit. As a matter of fact, he uses it to help automate his home and appliances! From exploring what goes into consulting customers on cloud products, to the nightmare of having your kids hijacking your Alexa with an endless stream of children’s music, this episode features twists and turns, leaving no stone unturned.Show Highlights:(0:00) Intro(0:40) Chronosphere sponsor read(1:14) The responsibilities of a Senior Cloud Engineer at Defiance Digital(2:07) Cloud product consulting(3:27) The challenges of working with Kubernetes(7:50) Mike's problems with AI(9:33) Challenges with home automation(15:38) Chronosphere sponsor read(16:13) The joys of home automation(18:34) Prefered hardware for home automation(20:10) Home automation and the impact on your relationships and kids(23:43) Going from teaching kids to the world of tech(28:42) Where you can find more from MikeAbout Mike GrayMike Gray is a technologist, currently employed as a Senior Cloud Engineer, with a focus on Amazon Web Services and Google Cloud Platform.In previous roles, he has worked with companies of every size, from single-digit employee startups to Fortune 500 companies. In a past life, Mike has worked as a professional musician and music educator.Mike is also an active open source contributor, splitting time between OpenVoiceOS and Neon AI. Think of it as open source Alexa, but all your data stays at home.LinksMike's website: https://graywind.orgMike’s email: [e-mail]’s Twitter: https://x.com/saxmanmikeSponsorChronosphere: https://chronosphere.io/?utm_source=duckbill-group&utm_medium=podcast

I'm now getting to the stage of my home automation journey where I can just about do a red alert automation. Welcome to Screaming in the Cloud. I'm Corey Quinn. Every time I travel somewhere, I try and throw a drink up. Earlier this year, I did that when I was in New York for the New York Summit, which I thought was going to be an AWS event and instead just turned into people from Amazon opining on Gen AI

and literally nothing else. When I was there, I met Mike Gray, who's a senior cloud engineer at Defiance Digital. Mike, thank you for joining me. Thank you for having me. I'm excited to be here. Complicated environments lead to limited insight. This means many businesses are flying blind instead of using their observability data to make decisions,

and engineering teams struggle to remediate quickly as they sift through piles of unnecessary data. That's why Chronosphere is on a mission to help you take back control with end-to-end visibility and centralized governance to choose and harness the most useful data. See why Chronosphere was named a leader in the 2024 Gartner Magic Quadrant for Observability platforms at chronosphere.io. It's always fun. It's because this is the literal answer like, oh, why do you have this guest on? We met at a bar. It's true. But there was a much more nuanced series of conversations that we wound up having, what, three months ago at this point because time is speeding up and, frankly, I don't like it. Let's start at the very beginning. What's your day job?

My day job as a senior cloud engineer at Defiance Digital, we're an MSP, but we function a lot more like consultants for startups. We help startups with their AWS problems. We do a lot of advising about architecture. We build most of the things for our customers in the CDK and then manage them because, honestly, because it's fastest for us. And then they can move as quickly as they would like to move and not have to worry about it or think about it. So it's a lot of CDK. It's a lot of architecture. It's a lot of diving into weird AWS nuance. And frankly, it's a lot of Django most of the time, which is weird.

I wouldn't have expected that. Are you one of those consultancies that's very tool forward as far as, oh, we solve things with the AWS CDK and that is how you go to market with it? Or does that just end up being the solution to a variety of problems articulated differently? It's not a tool focused organization. In fact, I've had that conversation internally because I thought, well, we're using a lot of CDK. Is this the thing we want to push? Is it a thing we want to market?

And the answer to that is no. We have a lot of CDK expertise in-house, including Matthew Bonig, who you had on the podcast some time back. And, of course, he's going to mention CDK as often as he can because he's very good at it and he's very good at teaching people. And honestly, it's a great tool for a lot of what we use. But we're not focused specifically on the tool. It's not part of our marketing, just happens to be a good use case for a lot of what we do. It's dangerous at some point when you start getting very attenuated to a technology and very focused in on it, you start to see it as the magic hammer that solves all problems. And I'm no more immune

to this than anyone else. Someone asked me, oh, how would I go about solving some computer problem? I'm initially thinking, well, okay, what is the AWS approach I would take to this without qualifying that is this on AWS? Because brace yourself, believe it or not, some people are not running something on AWS. I am as surprised as anyone to learn that, but apparently it happens. Other things that you can do, for example, a lot of the home stuff that I'm working on in my Kubernetes in the spare room would be horrifyingly expensive in a cloud provider. It doesn't belong there. Honestly, I would argue it doesn't belong on Kubernetes and Kubernetes doesn't belong

in my home, but those are ships that have already sailed at some point. It feels like there's a exactly there's a there's a love of a technology and people sometimes like to focus on that long past its point of utility and choosing the right tool for the jobs tricky yeah absolutely i want to clarify did you use kubernetes in the singular a moment ago i did it's running on some raspberries pie which is also pluralized oddly and the reason i did that is I lost a basic five-year bet with the internet that no one would care about Kubernetes in five years that expired in January. So I had to build one and give a talk on it, which I call my Kubernetes. Terrible idea in Kubernetes was the talk title. And I basically pointed out the problems that I had. Yeah, that makes a lot of sense. There's a number of my colleagues that are getting into Kubernetes the last couple of years, and we like to call it getting Cooper sweaty. Yeah, that's a good way of doing it, especially when something breaks and you're running something mission critical on top of it. It's like, how am I ever going to get this back up? Usually the storage system. And it always does come back, but it's touch and go sometimes. Absolutely. And there's a lot of cool tools for in-house. I run a Kubernetes cluster in-house.

I guess we'll do that. We're starting a thing here. And it's awesome. It just works. It's very cool until it doesn't. And then you have a real nightmare. And I understand now 100%. And I always advise people, let the cloud provider manage your control plane where possible.

Now that I've had a control plane break on me, I strongly advise people, please, please don't do this yourself. You really don't want to do this if you don't have to. It's similar to another hobby of mine. I've been doing a lot of work with a 3D printer for the last year and a half. And it turns out it's great when you get a 3D printer because there are so many things you can print,

specifically parts for your 3D printer. It becomes a self-reinforcing problem. And I look in the all namespaces, kubeControl or Kubectl or whatever you want to call it, list all of my running pods, and so many of them are just for the existence of Kubernetes itself before it ever gets to anything that would resemble a customer-like workload. Oh, yeah. I've got at least a dozen, two dozen pods that are just for the cluster, and then maybe three in my home cluster that do anything besides that. Precisely. And then it's, I've outscaled it too much, 10 nodes. Why? Because I make poor life choices. We've been over this. And that is significant

overkill. It's more complexity, and I don't love doing it. That also does not mean it's the wrong tool for other jobs, but it also probably isn't something that I should recommend as a solution to someone's problem when their problem description is not even fully out of their mouth. Sure, absolutely. One of the really interesting things, we've got a couple of clients today at Defiance that use Kubernetes clusters, and primarily what they serve out of Kubernetes is Django applications, which is the most bizarre thing I think I've ever run into because never in my life would I have thought, oh, I'm running Django. I think I've ever run into because never in my life would I have thought, oh, I'm running Django. I should probably put this, first of all, in a

container, period. That wouldn't have crossed my mind. But then to take those containers and hoist them over into a Kubernetes cluster, we've run into some very interesting issues from that use case alone. So many frameworks, systems, tools, applications, et cetera, were never designed with an idea toward, oh, this is going to be completely stateless, and then it can blow away and be recreated on a whim. Lots of things don't react so well to that. Oh, yeah. I mean, I love the tools that do. I love the applications that you can do that. It's like, yeah, let's spin up 100 of these a day. There'll be two of them running at any given time,

but they keep going down and up, and they die, and they come back. And it's great fun, but most applications, to your point, they can't do that. And when you try to make them do that, they break in just the most novel ways. I still have nightmare sweats waking up remembering some early client applications I smacked into in my career where you had a cluster of web servers or job servers running, and when a new one joined or one of them went away, all of the other servers needed to be updated to reflect the cluster constituency. So you don't have a single point of failure, you have as many single points of failure as there are cluster members, because if one of them goes down, the entire cluster would break. Yeah, that sounds like a middle school word

problem that I just don't even want to think about. It might have been one and I've just blocked it, but all of that sounds bad. So speaking of the wrong tool for the job, when you have a hammer, every tool looks like hours of fun. What's your take on AI these days? That was one of the big problems we had with the AWS Summit in New York is that it was about AI and literally nothing else. And that's great. AI is cool. AI has a value, but I have not yet found that it can solve every problem in my life. AI, my take on AI, how do I want to put this?

I think AI is very cool. There's a lot of problems that it solves. Most of the problems that it does solve are not the problems people are trying to solve with it. And I think there's probably a lot of problems that AI could solve that a lot of people aren't trying to solve yet. You know, you look at issues that we've had with machine learning, with natural language processing. There's a lot of those things that you can just feed to a large language model, and it's complete overkill. It's sandblasting a soup cracker. But it works, and it solved those

problems. So even if you've got a fairly small model, 3 billion parameters small, what a world we live in, it can probably solve a lot of those normal ML problems we've been having for the past decade plus. And I don't see a lot of people trying that or doing that. I like AI. I'm using it pretty much every day for one thing or another. It's great at doing tedious scripts that I don't really want to write. I don't get any enjoyment out of that. There's value in it for work, but it's not fun. So I can make the AI do it and it doesn't complain, or at least it doesn't tell me that it complains. It might be complaining under the hood and then there's a skim that gets applied to that.

And then it says, everything's great. I'm very happy to do this for you. I'm even running models on my MacBook. I'm running models on a tower in my house and using that. There's a lot of great things you can do with it. It just doesn't need to be everything that you do. You are a contributor to Open Voice OS, as well as Neon AI. Tell me about these projects, because these are not the type of project I would expect an AI skeptic to be working in. And again, you don't sound like an AI skeptic. You sound like most people I talk to where, yes, there's value to it, but shut up about it already.

There's more to life than this. Open Voice OS, also known as Ovos and Neon AI are companies that have carried the torch from a company. Some people might remember Mycroft.ai. They created this little thing here if you're watching the video i'm pointing to a cute little anthropomorphic voice assistant and the whole idea is that it's basically open source alexa you can't say that because amazon will sue the pants off of you because they don't have enough money and i got

into it because i had a few echo dots at home And I was running that and I was having it manage my home assistant, which manages all my lights and stuff in the house. And I didn't want to keep sending that information to Amazon. It made me uncomfortable. It made my wife uncomfortable. It made my father-in-law incredibly uncomfortable to come over and say anything at the house with those things. So I've been able to take that and replace all the Alexas that I had in my house. So everything stays in the home. It's not all on a single device. I think that's a mistake a lot of people have tried to make where you say, I'm going to run everything

for AI on one Raspberry Pi. Okay, have fun with that. There's going to be a lot of problems that arise from doing that, but you can run it all in your home, on your network. You get an old gaming laptop, and now you've got an NVIDIA card that you can use. You don't need the latest and greatest NVIDIA card to do a lot of the things that you would do. You need some text-to-speech. Use Piper or Coqui. Those are open-source alternatives, not super heavyweight to run. Coqui, you really do want to run with the GPU if you want to get the better quality voices. Piper, you get pretty good quality. You can run it on a Raspberry Pi. But then speech-to-text is where it gets complicated.

I'm rambling now. Feel free to cut me off. No, this is germane to my interest. I have something like eight of them floating around my house, and they are getting increasingly bad. It didn't used to be the case, but now I'll ask the one right in front of me

to play a song or set a timer, ask the one right in front of me to play a song or set a timer and the one upstairs, three rooms away, will instead respond and start doing it. It's bizarre. For something that is this invasive, I get a little annoyed with it. And especially given that the list of what I do with it is pretty small. I use it like you do, with home assistant integration to turn lights on and off. And I use it to set timers and my kids use it to play music. As long as I can nail those three use cases, I don't really care about anything else. The tricky one out of that, you know, you got two out of three out of the box easy. The more challenging part is music because the music industry, much like AWS, doesn't

have enough money. So they need to make sure that they have exclusive deals. And Spotify actually has it in their terms of service that if you hook a voice assistant to Spotify, you've violated the terms of service. They can terminate your account. So you cannot talk to Spotify in ways that they have not pre-approved. What have they pre-approved? Alexa, Google Home, things like that. There are ways around that. Do you still have your old MP3 collection? I know you had an MP3 collection at some point. I have it somewhere in an old backup drive, but yeah, I wound up uploading all of it to Apple Music years ago. Yeah, absolutely. Most people did. Most people have not run their own collection at home in probably 15 years or more. I was the same way. And I started looking around

and was like, well, if they won't let me do it with the streaming services, maybe I can play my own music because it's my music. Well, I think all of us with an MP3 collection can make a dubious claim to it being their music. But yeah, I digress. I bought the good stuff on CDs and I just format shifted it. And if someone has a problem with that, I really don't care, to be perfectly honest. Sure. Well, and with you on that, I'm more thinking about like the Napster days and we just won't talk about that.

We'll just skim over that part. So there's a couple of open source options you can do with your old libraries that still work and they're compatible with iTunes format, which, let's be honest, was probably the last format

that you had before you shoved it on that backup drive and put it in a box. So you can use a tool called Plex and some people don't like Plex. And some people don't like Plex because it's not open source enough. That's one of the fun things about the open source community, which I think you're familiar with that. If you don't pass the open source purity test, they're going to beat you with the magic hammer. Yep, always. It's instead of

they're the case study in letting the perfect be the enemy of the good. Yeah, absolutely. One reason I really like working with Ovos and Neon.ai is that they believe in it strongly, but they don't feel the need to beat me with that belief. So it's a lot more fun than the ones that like to flagellate you. I don't think I need to name names. We all have something in mind as we're hearing this. We absolutely do. It's one of the things I liked about the home assistance ecosystem that I've gotten, I've fallen down the well on is that even though I'm not a member participating in, I've just consume it. Almost everything I can imagine. And some things I

didn't realize are available in there. Like it shows up. My wife's Xbox pops up on the thing. Hey, do you want to manage this? It's like, not if I want to live to see morning, I don't, because if that thing stops working for an instant, she will murder me. The printer, do you want to manage the printer? It's like, buddy, no one wants to manage the printer. That thing is the devil. And so on and so forth. But there's a sense of it meets you where you are.

That's refreshing to see in an open source project. Instead of demanding I agree to a bunch of other terms and conditions, download a bunch of other nonsense, or in the case of Amazon, how do we make more money from this? and does the thing quietly, and fingers crossed, everything's working out so far. I can rebuild that pretty quickly, I think, if I need to. I would probably do that over trying to fix it any day, because fixing a Linux print server is not a good time. Oh, you have to be pretty deep in your cups for that to make sense. Oh, man. That's rough. Complicated environments lead to limited insight.

This means many businesses are flying blind instead of using their observability data to make decisions, and engineering teams struggle to remediate quickly as they sift through piles of unnecessary data. That's why Chronosphere is on a mission to help you take back control with end-to-end visibility and centralized governance to choose and harness the most useful data. See why Chronosphere was named a leader in the 2024 Gartner Magic Quadrant for Observability platforms at chronosphere.io. Yeah, but no, I just don't want Home Assistant to manage it because there's nothing I want to automate around that.

Oh, when I turn on the lights at two in the morning, start printing like an old-timey fax machine. Yeah, for some reason, that's not high on my list of home automations although now that i say that out loud be an amazing thing to do to someone else oh absolutely i'm now getting to the stage of my home automation journey where i can just about do a red alert automation where i just say red alert to whatever voice assistant i have and it'll change all the colors of my lights, and then you'll hear the Star Trek clacks, and that's the level of geek I'm inspired And it releases the Roombas.

Yeah, I only have one Roomba, but that's probably enough. It's scary enough. You could probably fix that, too. Fix it by adding more Roombas? Adding more Roombas, exactly. Then you could do Roomba jousting. That's true. Well, if anybody's interested in letting go of their old Roomba, they're looking for a good home,

I'll adopt it. I'll adopt a fleet of them. I'll take a video and share it with Corey and he'll share it with you. If I shared it with you, you probably would never see it, but Corey's got a good reach. Hopefully anyway, every time I do one of these episodes where I start getting deep into the weeds about home automation, I worry I'm either going to have an audience contraction or expansion and I never know which it's going to be. And I'm not brave enough to check the analytics to find out. That's fair. Yeah. So I really enjoy it. I enjoy it in that masochistic, I'm doing too much work at home for my hobby kind of way. But I'm really interested to see where this can go. You know, once we get over the basics, once we get it functioning a lot more like your random Echo

or your Google Home, what can you do with it from there? Because I think there's a lot more like your random Echo or your Google Home. What can you do with it from there? Because I think there's a lot of things that the folks at Amazon and the folks at Google and whoever else has been building these, they figured out how to do, and they were actually not that hard to do. But the problem with it is they know nobody trusts them with the level of personal detail and the level of private information that it would take to actually do it successfully. Well, everything runs in my house. I've got speech to text in my house. All the recordings stay at home. All the text to speech happens at home. Nobody gets the text that I want my assistant to speak to me. That stays at the house. So with everything here, what could I be doing that

people just haven't because they can't or won't trust the providers of those voice assistants. I think that's an interesting problem not enough people are spending time on, and rightfully so. A lot of us don't want to take work home, but what could we do with that? What do you use for the actual microphone slash speaker output? Do you just repurpose and reflash some of the old Echos, or do you use open hardware or something else? It's open hardware for the most part. There's different things that you can buy for Raspberry Pis. There are some older systems. A lot of it comes down to what is not a completely terrible microphone that's also not incredibly expensive.

One of the things that I'm doing with Neon that I think is most compelling right now is helping them create a hub system. So rather than having a whole bunch of high-powered devices in your home, it's a single server that does most of the processing for you. And then you can run the rest of it on anything you want. You can run it on a Pi Zero if you want, because it's not doing a whole lot. It's shipping all of it off to the central server, which, by the way, that's what Alexa's doing.

That's what Google Home's doing. So why are we trying to do something much more high-powered at home? You know, take the lesson and bring it in-house. It means your internet can go out and suddenly your lights continue to work. That's happened. I live in Texas, so power grid's really, really good. That was sarcasm. And when the power goes out, I can actually still run the lights because I've got it on

battery backup and then I've got a whole home generator. And again, that's because i live in texas and when the internet's out i can still run my house when the power goes out i can run my house enough to shut down the important systems before i completely lose the power um and i think it's kind of fun it's kind of cool you know if it's down i can still talk to these things and have them do things for me at home which i don't know it's a nerdy kind of fun a lot of people listening are probably like cool man sounds good but you know it's enjoyable part of it people are they're really into this or they're really not like some wit years ago observed that when you have a partner like i have a wife and you have iot or smart home

stuff one of you is really into the hobby and the other one is convinced that the house is haunted. And when things that they're trying to do don't work, they have opinions that they it wants to, although it does still happen. Bless her heart for being able to tolerate that. But she's seen enough success with it. She'll say, hey, can you do this thing? Can you do this other thing? Can we make this thing happen? And that's a lot of fun. I appreciate that she supports the hobby enough and I don't suck at it enough that she wants me to keep doing it. The hard part for me would be, I think, getting the kids to wind up adapting to change. I mean, kids are flexible to some extent, but yelling at Alexa is more or less how they make music happen.

And whenever you change that, for example, we'll be traveling and they won't understand that Siri, for example, will play music, but it will not respond to Alexa. And they get very angry because Alexa is the default name of the robot lady who makes good things happen yeah i've encountered that as well i've

got a eight-year-old and my son will be four next week and they have very much the same sort of response to it they've gotten pretty good they'll still call them the one name of the one that they remember um but they they're like oh that's right for this one. I have to say something a little different. They are somewhat flexible. Yeah. My daughters are seven and four and two weeks. And that makes it relatively easy from an MP3 perspective. Cause all I have to do is put the three songs that they listen to on repeat

forever on this thing. And we're basically done. Absolutely. They have completely destroyed my year in review on all of the music services. I look at this and I like, I know I'm not listening to this because I would have lost my mind. I, it looks deranged. Yeah. Especially when you see how many times it played that year. Oh God. Yes. I feel like I've listened to this more than the people who produce the track. Most likely. Yeah, absolutely. I'm working on a skill right now for my kids. They've

gotten really into some of these kids' podcasts. Go Kid Go has got a few that are really good. There's one out of Boston called Circle Round that tells a lot of folk tales, and they'll have some music going on. And they've got this whole touring production. It's an NPR thing. And kids love it. They want to listen to it all the time. Their favorite right now is The Adventures of Red Knight, which is a children's bedtime story gone wild.

It started off in the magical kingdom of kingdom and from it's not just knights and dragons and castles and princesses and what have you there's also a handsome spaceship captain who is a ghost who's inhabited the body of the local cat and then there's ray guns and plasma guns and just the sort of things kids love you know the the type of manic that children really are. Yeah, I find that mine are getting their fix on that from a couple of YouTube series that we curated and let them watch. But it's they're starting to speak with almost the YouTube influencer accent, which is really annoying. And it sounds like this and everything ends with a rise because it basically draws out and draws out engagement because it like they're asking a question and they're about to lead to a point, but they never do. And I, God, I can't do that anymore. I want to smack me.

I can only imagine what people listening would think. Sure. Maybe you edit that part out. Maybe not. I don't know. It could be fun. Maybe it becomes my whole new brand. Absolutely.

That's why I don't check the analytics. That's fair. Yeah, it takes me back to when I was teaching. I used to be an elementary school music teacher, and the kids talk like this until you do it back, and then they find it really annoying for some reason, and it goes away. You relatively recently entered TAC only about six years ago or so. Before that, as you said, you were an elementary school music teacher. What led you here? Kind of a strange road. Not that long. I mean,

it was 2016, a little more than six years, not much. And so I'd been teaching and enjoying teaching, loved working with the kids, had a lot of success and was able to get VH1 Save the Music grants for the school. Some other grants got a lot of instruments in, concerts, families were happy type of thing. But I don't know if you have any teacher friends or family, but it's kind of a demanding and punishing profession. I like to say that teaching is a survivor's game more than anything. That is a massive understatement. Every teacher I know is more or less a living saint. Yeah.

Well, I wasn't. I had a hard time handling some of the demands of particularly administrative type of stuff. You know, not so much with the kids, although that that was a lot. But a couple of events happened that really led me to kind of rethink my life. One, my son was born. So that was a big change. First child. Then you go, oh, wow, life has changed a whole lot.

Let me think about what that change means for me. And then unfortunately, 10 days later, my dad passed away. So sorry. Thank you. Two massive life changes happening right one after another. I had a long think, sat down and had that think. And I thought, well, you know, this has been kind of rough and it's not especially lucrative. And I've got a family to worry about now. It's not just me and my wife, but you know, I've got, I've got to worry about the child too. And I wonder if I could do better. I wonder if I could do anything else. So I went to a free code camp.com. Yes. Somebody who's done that has actually gotten a job in tech. I'm hello. Nice to meet you. And I reached out to my network. This, this was a lot more successful than doing free code camp was, if I'm being perfectly

candid. I reached out to the network and I said, I'm interested in getting into tech. I don't think it's going to happen today because I don't know anything. But what skill sets are you looking for in new hires? You know, if you're hiring somebody junior, what do you want them to know? And the best response that I got was from the person who actually ended up hiring me eventually. He said, well, we look for experience. I said, great, that's not helpful. Can you give me a little bit more to go off of? So then he gave me details. He said, well, go

figure out how to code this. And you need to understand a little bit of networking and you need to probably figure out what the cloud is. Because for me, clouds were the fluffy things in the sky, which they are, but they have a lot more meaning now. We're screening in one now, if I'm not mistaken. Indeed, we are. Yeah, I spent some time working on those things. And he kept up with my progress. And he went, oh, wow, he's actually doing that. I don't know if you've ever had somebody ask you, hey, how do I get into tech? You give them some advice, and then they don't take it. I usually warn them.

Yeah, that's also fair. But you're right. People do ask for advice when in some cases they're looking for either an easy path or looking for validation of a decision they've already made. Not most people I found are looking for advice, aren't really looking for advice. I'm always taken aback when someone chases me down and says, yeah, you gave me some advice a few years ago and I followed it. It's like, oh, no, what did I suggest you do?

I hope you're not destitute. I'm so sorry. Yeah, exactly. But you went from being an elementary school music teacher. You had a couple of big life changes and then one thing led to another. And you're working on Kubernetes. That sounds like a destructive spiral, if I'm being perfectly honest. Well, I mean, I certainly have aged in that time.

The nice thing about it is that tech is much more lucrative. So that makes a lot of things easier to bear. And on top of that, you know, teaching was incredibly rewarding. There were some really amazing highs and some absolutely devastating lows. You know, my range was like this, you know, really, really low, really, really high. In tech, it's a lot more narrow, and that's okay. I don't need my emotional highs and emotional lows in life to be from my job. In fact, I prefer they not be. I'd like the job to be a little bit more steady. I still want it to be fun. I still want to enjoy what I'm doing and the people that I'm working with, but I don't need it

to be thrilling. I've had the thrill. Thank you. It was kind of terrible. I'm good not reaching those heights. I'll find other ways to get there. A job, at least for me, has to be something that I can leave at the office and go home. Otherwise, it becomes this all-consuming thing. Well, I already have family responsibilities and things that consume me and roles I need to play in my life where I don't get to be all work all the time. Believe it or not, my conversations at the dinner table are not entirely like my Twitter feed. Not entirely. There are some one-liners, I will admit. Sarcasm is our first language spoken at home. That's fair. And it's important to be bilingual, at least at home,

if you can. Exactly. I really want to thank you for taking the time to speak with me about all this. If people want to learn more, where's the best place really want to thank you for taking the time to speak with me about all this. If people want to learn more, where's the best place for them to track you down? Probably email because I don't maintain a social media presence, feel much better about that. Now that I'm not on social media, my life has improved quite a bit. Email me, find me on Twitter, I guess. I don't really use it. So I'd probably say, how did you find me? Email is going to be the best if you really want to continue chatting about these things. Or do what I did and just show up at a bar. That's very likely. Yeah, that could work. Thank you so much for taking the time to speak with me today. I really appreciate it. Thank you. It was fun. Mike Gray, senior cloud engineer at

Defiance Digital. I'm cloud economist Corey Quinn, and this is Screaming in the Cloud. If you enjoyed this podcast, please leave a five-star review on your podcast platform of Thanks for listening to the podcast.