Backend Banter

# Melkey

**Melkey:** [00:00:00] honestly just calling a spade a spade. I've never seen a check or rent being paid with passion. The great thing about software engineering is you can make a lot of money, like a disgusting amount of money. FAANG jobs, we can pull on levels.Fyi like junior developers are making 200,000 total compensation which is.

un like, that's more than both my parents combined

**Lane:** Yeah. That's like a mechanical engineer with 30 years of experience or

**Melkey:** more yeah, that's public. Two with 30, like a professional engineer, like with, like you're not gonna get that.

​

**Lane:** By popular demand. I am here with Milky. Welcome, man. How's it going?

**Melkey:** What's up man? Appreciate the opportunity to have me here. Excited. I know I have to reschedule a few times cuz I'm super bad with planning, dude. I'm like, I just can't plan. It's such a bad thing. I go with the win, but I'm excited.

**Lane:** So I'm gonna have to leak this. I've learned that[00:01:00] I'm now stereotyping all Twitch streamers as bad planners.

**Melkey:** Yeah.

**Lane:** I think there's a pattern, now, obviously, every single one of my listeners probably already knows who you are, but for the 5% that isn't aware of your fame go ahead and tell us just a little bit about, who you are, what you do for work and all that kind of stuff.

**Melkey:** Yeah, for sure. Okay it's fitting backend banter. I'm a backend engineer. I'm a backend engineer that wears a front end mask, I would say. But yeah, my name is Milky. It used to be Milky Dev, but it removed the dev part.

And I work for Twitch right now. I'm a ML infrastructure engineer at Twitch, so it's like infrastructure as code, lots of backend stuff, lots of services that I own and create. And I stream on Twitch, so you catch me there. We talk about a lot of different things. Try to keep up with, the latest innovation in the tech world and specifically in the front end world.

That's why I wear that mask, because professionally, I get paid to write code for. Backend stuff, an infrastructure pipelines, [00:02:00] all that greasy, heavy level, deep dive stuff, but yeah. Yeah, exactly. But I think a lot of my passion, a lot of the apps I build are more frontend centric, so I definitely have a good feather in my cap when it comes to understanding the current changes in both worlds.

**Lane:** Cool. So we are completely opposite then, because today for my job I was forced to write some CSS and now I want to kill myself. And I'm excited to, I'm excited to talk about some backend stuff. So it sounds like your, correct me if I'm wrong your role is almost like a data engineer.

You're like building tools for. For ML stuff.

**Melkey:** I wouldn't necessarily say it's a data engineer. There isn't a lot of tools I build have a sincere focus with handling data and providing data at large scale, like at Twitch scale. And so we, I've built tools that, provide data for models to be inferenced and trained on. And that is a part of my job, but a lot, the majority of it is more the infrastructure and setting up, how do we even create the [00:03:00] pipeline to process this data?

How do we scale it? What do we use to even, what's the first step? How do you go from handling data that's, let's say, 30 a stream of 30 data points a second to handling a million data points every minute or something like that, right? The scale is much different. You can't really use the same infrastructure.

And it's also How do we apply the smart infrastructure Because you don't wanna just throw the bag at it and just like more money will solve your problems, you have to be smart about it. There's always cost implications. Even though I work at Twitch, which is like under a w s, we still have to be mindful of our cost there.

So it's a lot of stuff like that. And, there's a lot of backend core service of service stuff that I also get my hands dirty on.

**Lane:** Got it. Okay, cool. So we, we have this I would argue a very similar background. I, before working on boot dev My title was backend engineer, but I did a lot of this data engineering stuff. The data I worked on, though I'm sure is quite different. We basically ingested a large portion of Twitter and Reddit and text data and did a bunch of N L P text analysis, but at the end of the [00:04:00] day, I was basically just ingesting all of this data, cleaning it up, shoving it into some database, and then, later we do some ML stuff analysis and aggregations and

**Melkey:** Yeah, that's super similar to the stuff I do. The data might be different, but the underlying gold sounds the same.

**Lane:** Yeah, so do you work mostly with like video, like encoded data or are you doing like likes and subscribes and like that kind of event stuff?

**Melkey:** Yeah, so there's a separate team altogether that handles everything, data. I don't touch any da core data ingestion stuff that's, like I said, like a whole separate team. My stuff is more so like I. Yeah, like every, everything else when it comes to a user interaction, a Twitch platform. So when they click something, when they subscribe, when they unsubscribe and they follow, when they unfall any action, all of that gets piped in real time into like our data lake, into our streaming platform and all of that stuff.

And then it's how do we process the scale of that per user, per channel, per user channel pair. And it, it gets. You can imagine for one channel, [00:05:00] let's say a channel like x QC pops up, right? And X QC not only has 30 K live viewers, but let's say 10,000 of them are constantly talking in chat.

So we're ingesting all of that data and all those parameters. And then let's say from that 30,000, let's say at a time, every second a hundred are gifting a sub or in the process of resubscribing. And from that 30,000, let's say 5,000 clicking follow for the first time. Or another one is, let's say add those 30,000 10,000, have another tab open with another streamer, and then we do the exact same thing all over again, all matching that to the same user and all their channel pairs.

So it gets pretty gnarly pretty quickly.

**Lane:** 30,000 concurrent users. That's like half of your current viewership, right?

**Melkey:** Yeah. Yeah, that's, yeah. Roughly. Yeah. Actually, she's, he's coming from my title. I think I'm still quite a, a ways ahead of him. But, he's on my radar. He's on my

**Lane:** got a healthy lead. You just

**Melkey:** yeah. For now.

**Lane:** Cool. Okay, so this is actually [00:06:00] super interesting. I didn't realize this. So your users, like the users of the, like data that you're aggregating and the infrastructure you're building, like it sounds to me like that's basic. Is that like the product marketing team, the product team, are they going and like trying to figure out, What, like what viewers and streamers are doing and how they're using the products so they can improve it, or how does that work?

**Melkey:** Yeah, so every org in Twitch has a product division. My org is specifically focused on how to make money. It's like the commerce org and obviously with that we try to release new products or improve existing products. So I work very closely with the product stuff on Twitch, building new products, like you said.

But I also do some other stuff, which is like fraud. Detection. So I work with a lot of high level applied scientists that are trying to in investigate and detect which accounts are fraudulent or doing fraudulent transactions, and I'm building the service around that to action on, if that user is fraud, what do we do?

**Lane:** Got it. [00:07:00] Okay. Yeah, that sounds. Sounds really fun. What kind of technologies are you using from a high level to, to do this stuff?

**Melkey:** So we're fully on aws, so all of our infra is on aws. So that could be like, step functions. That could be s3 our streaming services, all of that. That's all done through aws. Any service that AWS offers, that's us.

**Lane:** You use as many managed services as you can?

**Melkey:** Yeah, we try to,

**Lane:** bare bones, VPCs or whatever.

**Melkey:** We do a lot of bare bone VPC stuff as well. Like we talk, like we get into the networks, like I've had to configure stuff from security networks, from the private vpc. Like I had to do like pretty gnarly, low level stuff like that too. That's not my favorite. That stuff I goes over my head.

I'm not too well-versed in that area. I'm, I can get around. But yeah, we actually do a lot of like managed services. We try to like, Airflow stuff, managed M W A, all that cool stuff. We try to use as much out of the box with changing the parameters to fit our needs. And [00:08:00] then the majority of our code is written in go So goaling for like a backend services?

**Lane:** Yes. Oh God. That's awesome. Okay,

**Melkey:** Yes. So we're majority goaling and our, and we have a very tight coupling to type script for some infr deployment stuff, cuz we use type safety for, deploying our CDK stack and all that.

**Lane:** Amazing. I love hearing about other people using Go in production. I don't know if you knew, so Boot Dev is a place for people to learn backend development. That's, this podcast is Center Out, obviously backend development and we use Python and Go. So I'm always super excited,

**Melkey:** Yeah. Writing a backend in Python is a little eyebrow racing. I won't lie to you.

**Lane:** I don't do it in production. We teach Python because, a lot of our students are learning programming for the first

So we bring you up in Python to teach you like basic coding skills. But then all of the backend related stuff, at least on boot dev is taught in go,

**Melkey:** Okay. And what languages, how did you, what language is boot written in?

**Lane:** go.

**Melkey:** Okay. All go.

**Lane:** So I've been a go developer for [00:09:00] almost my whole career,

**Melkey:** Oh, okay. Nice.

**Lane:** primarily and I did Python in school. I've done a little python in production. Not a ton, obviously tons of JavaScript. You can't get away from that. But no.

**Melkey:** front for the client interfacing parts, right?

**Lane:** yeah, exactly. Don't write it on the backend unless I have to.

**Melkey:** Yeah. I avoid I wouldn't say I avoid JavaScript on the backend, but I. A lot of the apps I build now, just for sake of convenience, like with Next and Next Auth, are primarily just full fledged JavaScript applications. Just because what I like just building apps.

I like building apps that people use and. There's, there was one instance where I was making something and I still have it on the, on on my project roadmap. It's it was an app that allows like Twitch streamers to see their data in real time. So we use like the Twitch api, everything's it's not, because I work at Twitch, I get this data just like Twitch API stuff and the backend is written and go, cause I'm processing data, I'm gonna have to do some pretty crazy stuff with it.

I just prefer go, but, Using [00:10:00] Next auth and next is such a win. Especially when it's dealing with user databases cuz like handling cookies, protocols, auth stuff like with Just Go and then having your JavaScript application, like it can get messy super quick. And so I'm a huge fan of rolling my own OTH with something like Next Oth.

It, trying to get that, trying to have a JavaScript or a type two project with Next, which is technically a backend framework for type script with a true go backend let go. You can imagine it's pretty, pretty messy and hard to maintain fast. But I think I found a decent medium for it for right now.

**Lane:** Yeah, so that's actually like you've pretty much described the stack that I prefer for. A significant number of web apps, which is basically like this full stack kind of type script backend front end for like the application itself. You know how there's like the backend for the front end and then there's the actual

**Melkey:** Back in. Yeah. Yeah.

**Lane:** Yeah. So like the backend for the front end makes a lot of sense for me to be in type script. But then like times you need to do [00:11:00] like heavy data processing and if you hate money, then you can do that in type scripts. Some of us don't hate money,

**Melkey:** yeah. And just it it makes, there's not, the tooling isn't like robust enough I think in type script to handle true data processing at that scale. Go is go. That's what kind of a go is made for, right? It's it was built to be used as a microservice for these kind of purposes.

**Lane:** So now that I'm like super happy about your tech stack and we've geeked out about that, I wanna talk Yeah. I wanna talk a little bit about Your, like, how you view, how you got to where you are. So on your stream, a lot of times you talk about career stuff, do I need a CS degree to get to where I am today?

You're a, correct me if I'm wrong, senior developer at Twitch.

**Melkey:** yep. Yep.

**Lane:** Cool. How did you get there? Did you get a CS degree? Do you need a CS degree? Let's dive into that topic a little bit.

**Melkey:** Yeah I preach what I believe in always. I'm never gonna be a person who like says something, but I do something outside of [00:12:00] what I say and all my experience comes from what I've witnessed, right? So this may be different to other people. And because I've realized like my audience, a lot of my audience is like very European based which is cool.

Or I should say not even European based, but outside of, let's say America, like the United States. So there's like a lot of people in Brazil, a lot of people look just worldwide. For those people that wanna like actually move to United States, like I would just wanna say that getting a CS degree is the best thing you can do.

There's no alternative if you're in a different

**Lane:** mean move and get a CS degree?

**Melkey:** No, get a CS degree in your current country and then when you apply to a Visa to move your chances of getting approved for your visa drastically go higher if you have a, if you have a CS degree.

**Lane:** so like for Visa purposes, having a degree will make that whole process easier.

**Melkey:** Yeah, and a specific CS degree like that, not just a degree.

Like it has to be a degree at the very least in stem, like a stem CS degree. But if you're going, applying to Amazon and let's say you get to a point where they're [00:13:00] like ready to offer you a letter or something, but they have to sponsor you for a visa, like they probably won't continue that conversation unless you have a CS degree, especially if you're not in Canada or Mexico.

**Lane:** That's really interesting. I never even realized that, but it makes a ton of sense and I've never experienced the whole visa thing at all,

**Melkey:** yeah. So I'm, I have a visa, but my visa, because I'm Canadian. My visa is super easy. It's a work authorization visa. So I don't it's super easy for me to get this visa, if I just need an offer letter and they're like, yep, you're good. So that's it. But back to your, the root of your questions like, do I think people.

I guess in the United States or Canada, either a CSV to get a job and like absolutely not. I don't have a CS degree. I have a bachelor's in mechanical engineering. I have a master's in ai, which is very weird. A lot of people are like, how do you not have a CS degree if they have a Master's in ai?

So my master's is in applied science of engineering with ai. It's still not considered a CS degree and my Bachelor's of engineering is definitely not a CS degree. I [00:14:00] was able to get, like I went to school for mechanical engineering, I realized it was shit. Cuz you're gonna get paid pennies. The work sucks.

Everything about it sucked. And I realized this in my last semester my very last semester of my last year at school. And I was like, shit this is not good. Like I'm about to enter adulthood, hating my life. And I remember distinctly, my mom even reminds me about this all the time. She's I remember the time you came back home and you're like, Man, my life sucks.

I'm gonna get a mediocre job, get a mediocre pay, and live a mediocre life. And now that's like not the case. But yeah, what I did was I just learned I just learned programming. I just like, like on my last semester, I literally I wanted on programming and I knew I didn't have enough time.

So this was in February. I'm graduating in like end of April, so it's two months away. I'm like, I can't go back to get an underground in computer science. I don't have enough money for it. I had to pay, I had no, all my loans, all my grants were done. I had no money. So I'm like, what? What do I do? [00:15:00] And the only thing I could think of was like, okay, I have two months.

I have to meet a mentor. That's the only thing I could think of. So I went to my school's computer science lab. I found them somehow didn't have access to the building, so I had to wait until like class finished and I like went in

**Lane:** walk in when the

**Melkey:** Yeah, walked in. I was a student, all that. I didn't have access to that building.

And I specifically searched up this professor. He was a professor who's a 40, under 40 Graham Taylor. He earned his masters of AI under the godfather of ai, Jeffrey Hinton. And I go up to his office, it was on the fourth floor, and he's oh, you're here for like office hours. It just happened to be office hours, right?

And I'm like, no. I'm like, no. Like I, I wanna learn the program. And this guy, he's just not a bullshit guy's. What, but what I don't understand, he's I wanna like work. I wanna work in your lab. You've taught me this course before. Like in, during my undergrad, you probably don't remember me cause I sat in the back, but I wanted to program can I join your lab?

I'll do anything. I literally said I'll do any kind of work as long as I learned her to program. And he's yeah. He's I'm not gonna [00:16:00] pay you. I'm like, I don't wanna be paid. He's like, all right, cool. Start tomorrow. And again I still don't have access to a lab, so I had to sneak in there, went into his lab, didn't have access had to knock.

Someone opened it. And I just stood there until someone's who are you? And they're like, okay. I explained the situation. They're like, oh, okay. Yeah, professor Taylor told me about you. You're gonna parse CSV files. He sat me down on his desk, on a desk beside him, turned it on. Gave me a OneDrive dump of CSV files.

And the student was a PhD student, so he was doing his PhD and he needed like this data pared and it was like, bitch work. Okay. And he's go ahead and do this. And that's the only instruction I got. And I'm like, how? He's use Python and that's it.

**Lane:** Dude CSV parsing, I think is the CSB parsing is to backend development. What like building login forms is to front end development. Like it's just, you just have to do it so often.

**Melkey:** Yeah, exactly. [00:17:00] And it was really good experience. Like I l I introduced him to like Python, like pandas, imports, all that stuff. Took me a very long time and I didn't even get the data right to him, but I. I was in that lab all the time. All the time. Like literally I would be the first one in there when I got access eventually.

And I would just, I would eat lunch there. I would take my courses there, like online courses. I would just literally live in that lab and I would just continuously like parse CSVs. I would eavesdrop on like conversations cuz this was an AI lab. So they're talking about like neural networks and pie torch and TensorFlow.

And I would hear the word TensorFlow. And I'd be like, what the fuck? And I would just Google it. I'd be like, what's TensorFlow on my computer? Look at it. And I would use it. And that's eventually how I got into neural networks.

**Lane:** That's awesome. So how, like you, you have this background, a bachelor's degree in mechanical engineering, a master's degree in ai. Which like what? That you actually learned in school. I'm sure that a ton of [00:18:00] stuff you learned in school actually is applicable in some way to programming. What has been the most applicable stuff and what's the stuff where it's kinda been like, ah, that was a waste of time and money.

**Melkey:** Honestly, every course I took was just like a waste of money. None of that is applicable to anything I do day

**Lane:** Nothing. Not even the math courses.

**Melkey:** No, not even close. Like I'm not computing jacobian as a neural network, like TensorFlow and Pie Torch handled that for me. I'm not computing matrices or multiplying matric together.

I'm not doing any of that stuff. It was a fun ex little tidbit to understand how things work under the hood. But other than that, like that's not applicable. Cuz at the end of the day, business doesn't care about what's under the hood. Business care, the sound of when they slap the hood. So that's what they want.

And yeah, I would say another courses are very applicable. The only thing I'm grateful for with my master's is that it bought me time. And that time was more valuable than any skill I learned.

**Lane:** So while you were doing [00:19:00] the Master's, is that when you were learning programming?

**Melkey:** Yeah.

**Lane:** Oh, got it. Okay. So you were like actively getting a master's degree in something that you were like, I'm not, I don't know about this, like it's not gonna be useful. But at the same time you were just learning programming.

**Melkey:** Yeah, so that story with the lab was in February and I worked it all the way until end of April. I learned like a decent amount of Python and neural network stuff. Cause like I said, I was ea dropping and there's like applying that stuff and it was cool. People would show me how to do it. But then I graduated and I'm like shit,

**Lane:** You graduated with your master's degree.

**Melkey:** No, I've graduated my undergrad. At this point. I have yet to be, I have yet to join. I have yet to, I didn't know at this point. I'm still doing my undergrad. I did not have a master's lined up, nothing. I was just working in the lab of this neural network professor for free. And then I graduated. I was like, okay, I gotta find a job.

Somehow as a programmer, I was like, determined. I was like, I gotta find a job As a programmer, I was able to get like this random like consulting gig for a couple of months. It was trash. I like barely [00:20:00] talk about it. Like it was just so bad.

**Lane:** Like with a web dev company or something.

**Melkey:** Yeah, exactly. Just, yeah. It was like I was gonna pay like 40 K Canadian a year like noth, like it was just terrible.

And then

**Lane:** like that's a common trap, by the way, for college grad grads. I know a lot of people who like graduate and then quickly get a job making like a very low amount for like a company that just farms out web developers. And it's always I think it's an okay place to start, but do not stay there very long.

**Melkey:** Yeah, agreed. A hun it's an okay place to start if you can find an okay company that lets you grow. But I could see the writing in the wall immediately, so I'm like, this place is just not for me. And then, so I was like working there, hating it, trying to learn how to program. And then I got a, like an email from that professor and he is Hey, like we're rolling at a master's program.

We wanna find students. Do you want to do you want a full ride? And I'm like, yeah, I'll take the full ride. And so then I went back to school, got my master's.[00:21:00] And that is when I start, like I was doing my master's, had to build like my hy my thesis all while learning how to program.

**Lane:** So why get the Masters in AI at that point? Why not get it in CS or something?

**Melkey:** It was what was offered to me. It's cuz because I was working in the

**Lane:** it was just that program. Got it.

**Melkey:** yeah. It my, that professor had. Some sort of affiliation with that program, and so he could only offer that program. He couldn't offer the Cs I, I wasn't in a position to be choosing,

**Lane:** yeah. Yeah. Okay. That makes sense. Wow. Okay. I've never heard I've never heard anyone take quite as hard a stance as you have. I can think of, during my CS degree there were probably, I. Probably like a quarter of the classes, o of all the classes in my four years where I'd be like, that was actually useful.

And like I've used it at some point. Things like data structures and algorithms even like object oriented and functional programming, like understanding those concepts. I had a programming languages class that kind of broke down those paradigms. But like the other [00:22:00] three quarters, like I'm with you, it was just terrible.

I took an acting class because I needed to maintain my scholarship, like it was

**Melkey:** Yeah.

**Lane:** Just, it's just not,

**Melkey:** You just had to do it.

**Lane:** just had to do it. Yeah. It's that kind of stuff is like okay, when you are 18 years old and you're just like living super cheap and college is paid for again, cause I have a scholarship, but if you are in your late twenties or thirties and you're trying to make a shift into tech like that, that can be such a waste of time.

**Melkey:** Yeah, and that's exact, like especially for people who are making that shift, like you don't need a degree. That's a trap. Cuz like you said, people at that age, they have a lot more responsibility. Like they probably have a family. Their parents are older, they might have like their own child, whatever, right?

Like they have bills to pay, mortgages potentially. Like you can't fuck. I dunno if I can swear or not. Yeah, but you can't mess around and go to school for four a year to gain some dusty degree. Like at that point you definitely need to. Do it yourself like a D a CS degree at that point's not gonna [00:23:00] give you any more information cuz it's not only that cost a lot of money, it costs a lot of time and like time is worth more than money in, in, in certain a aspects, right?

So that's why when you ask what's the one thing I was like grateful for? Like it was the time that my masters bought me. I never talked about my thesis cuz I'm not really proud of my thesis work, but I'm proud of what I did in that time of basically learning how to program and getting hired as an engineer.

**Lane:** So we know what you wouldn't do again, right? If you were to reverse engineer your career at Twitch, so if you knew let's like have a thought experiment. You know, That your goal is to get your current job at Twitch and you have I don't know, four years to do it. So we're shipping you four years in the past.

How would you re-engineer from the ground up, that learning experience? Would you go to college? Would you not go to college? What would you do instead? What resources would you use?

**Melkey:** No, I would not go to college at all. That wouldn't even cross my mind. That would like absolutely not. What I would do, and this is assuming do I need to [00:24:00] work or anything, or I just am I like a college student?

**Lane:** Oh, yeah, that's a good question. How old are you now? Do you mind me asking?

**Melkey:** I'm 20. I just turned 26. My birthday was last week.

**Lane:** Let's say you're 22, and let's say 22 is right about the age. Let's say you graduated with Completely unrelated degree. A degree in cartography,

in map making. Okay and now you're like, oh crap. I actually, I don't know why I got this degree in making maps, but now I want to go get a job working at Twitch.

And the reason I'm using this scenario is actually, this is like a lot of the students I talked to on B Dev or in a similar situation, it's like just got a music degree. Oh crap. I actually wanna be a software engineer.

**Melkey:** Yeah. If in that's the position, there's two things I would first of all do. I would work enough as a cartographer where I can basically spend 90 days not working. I would, that was my main goal. Like before I even touch programming, I would maybe lightly like brush up on like certain things at night.

After I do my job, I have to do my thing. But I would save up [00:25:00] for 90 days. Or

**Lane:** 90 days of work free like sabbatical from

**Melkey:** Three months, basically. Three months. I would save for three months, right? And this is assuming I have a job. If I don't, then I'm going harder. In those 90 days, essentially the first step is buying 90 days, whatever, however you can do it.

If you have to get a loan to get 90 days, then do that. If you have to just work for, I don't know, four months to buy 90 days, I would do that. The goal number one is to buy time, however you can. And then the first thing I would do is lead code. I would lead code. Every single day. When I was, when I got my job at Twitch, I was le coding 12 hours a day every single day for I think it was like four weeks I stopped streaming.

I'd take this shit seriously, right? There's no shortcuts. So I was literally like le coding for 14 hours a day. All I did wake up le code until I went to bed. Like my eyes would, I would be in a basement. No, sun didn't go outside. Didn't

**Lane:** bloodshot eyes

**Melkey:** Yeah, pale.

**Lane:** binary tree for the

**Melkey:** Yeah, just pale skin, like everything like that man.

And I would [00:26:00] like smoke JUULs to make sure, like I'm like active, my brain's functioning. But yeah, I'd buy 90 time, I'd get 90 days and then I would lead code for at least eight hours a day of leak code. So I'd wake up,

**Lane:** the 90 days you'd delete code.

**Melkey:** yes easy. I'd wake up and I'd put eight hours every, like seven days.

There's no days off. I would take maybe Sunday days off, only Sunday would be the day that you can actually relax. So eight hours a day like that would be your nine to five. And then I would give myself two or three hours. So let's say that brings us to eight or nine. And then I'll do projects.

I'll build projects using the latest technology. So if I wanna be a front end dev, I'm spinning up T3 stack and I'm figuring out what's going on here. I'm watching Theo, I'm watching Prime, I'm doing whatever I can. Watching YouTubers build. Shitty ass projects and I'm mimicking it. I go watch ACY Media, he's doing a course in me Stack.

I'm learning Mest Stack. And that would be my evenings. That would be like from nine to midnight or eight to midnight.

**Lane:** I love it. Okay, cool. This is backend banter, so I have to [00:27:00] like, find points I agree with you on, and then also find points I disagree with you on so that so that it's fun. Okay. So grinding and working hard, like this is not talked about enough in software development. Like it is not easy to learn to code.

So I love it. Anytime I talk to somebody who's not sugarcoating it you're gonna have to write. An as load of code if you want to get a real job, right? Like people are hiring you to solve real problems. Not build to do apps. So I couldn't agree with you more there. Lead code.

That's an interesting take. I want to dive into that though. I love lead code. I think the idea of

**Melkey:** I hate it.

**Lane:** take yeah, it sounds like you like it more than me. Maybe.

**Melkey:** I'll explain why. I'll explain why, but I did test Lee Code.

**Lane:** Okay, so the way I think of LEAP code or the way I recommend students to use it is okay, I'm gonna go learn this concept.

So I like take a course and I learn, you know how, I don't know, big O notation works, right? How we design performance algorithms and then go grind leap code until I get it

Right. It's like putting it into practice. Is that how you think about lead code or [00:28:00] is it something else?

**Melkey:** No, not at all. I don't do LeEco to learn anything. I Lee Code is simply an avenue, it's like a road. And the reason why I would do LeEco is cuz you asked me Twitch specific,

**Lane:** Ah, okay. Yeah.

**Melkey:** Any fang job. And I'd recommend, honestly just calling a spade a spade. A lot of people like going to program cuz like, oh yeah, you get to solve problems and ooh, like you have a passion for program.

That's all good. I love that. I've never seen a check or rent being paid with passion. The great thing about software engineering is you can make a lot of money, like a disgusting amount of money. FAANG jobs, we can pull on levels.Fyi like junior developers are making 200,000 total compensation which is.

un that's more than both my parents combined

**Lane:** Yeah. That's like a mechanical engineer with 30 years of experience or

**Melkey:** more yeah, that's public. Two with 30, like a professional engineer, like with, like you're not gonna get that. So I, like at that point you want to build stuff and get paid handsomely for it. And the way you get handsomely paid by is by working at a big [00:29:00] tech company like a FANG company.

And there's actually another benefit that I'll get into for a new person going into Fang over a startup and lead code is the way to go into that. If you are able to prove, cuz you'd be going for a junior role and hope, like I hope people who are 35 or even older or maybe younger, who are switching their careers.

If you think you are too good for a junior role, you're not. Majority of people who are going outta these boot camps are making this shift in their life. You are probably gonna be a junior engineer if you're applying for the first time. Even if you do a bootcamp or even if you know how to spin up a complex app and use thread, it's like you're still a junior developer.

You can't really side project away out of that title. That works with like actually working with teams and enterprise applications and like getting stuff done. When you leak code, you'll leak code into gain an entry level position. That entry level position, that big tech company is gonna pay you a lot.

And the other benefit is when you go into a company like that, You are [00:30:00] not going to be able to write spaghetti code and when you work at a big tech company, you're gonna be working on bigger teams. Those bigger teams are, have more senior advocates, senior developers, and those seniors are gonna be able to guardrail your spaghetti code.

Yeah. The first you make a CR are a pr. It's gonna look like dookie dukes, but that's why you have the senior and the mentor who's gonna be telling you, okay. Instead of writing a for loop like this, do like this. Or instead of writing Nest if statements do this. And that's how you learn good practices while getting a great salary, while working for a good company.

All because you grinded Lee Code. So for me, I don't, if you ask me now can I do binary surgery? No, I can't. I don't even want to, I don't think it's embarrassing to say no I can't. Absolutely not. But I know that if you give me a week, I can code you the best binary surgery you can think of because I, if I need to learn it for a reason, I'll need to learn it.

So yeah, that's my stance on Lee Code and why I actually heavily advocated for junior engineers and [00:31:00] people getting into it. It's cuz it is the best kind of foot in the door for a great future.

**Lane:** So let me dig into this just a little bit deeper because I think there is a bit of nuance here. You are saying, if I'm understanding correctly, that the FAANG companies, like they will have rigorous technical interviews that quiz you on the leet code stuff, competitive programming type problems, algorithms, data structures.

How can you cleverly solve this technical challenge? So if you are interested in Fang, if you're interested in Twitch, you need to spend a disordinate amount of time grinding leet code. You actually do need to learn the concepts first, right? You do need to go learn why we use binary trees in the first place, but then you need to actually go solve some problems with it.

Do I need to grind leet code if I'm going to startups or mid-size companies?

**Melkey:** No.

**Lane:** What do you do instead?

**Melkey:** That's all project based and that's all like very catered project based, right? So if you're gonna be applying to a [00:32:00] startup, like something from YC or something like that and you are specifically targeting companies for a front end role. A lot of the times those companies explain the tech they use, they're, they don't hide it.

It's because they're looking for other people that have used that tech that they can onboard almost immediately and get stuff done. Those companies, for every interview I've had with a startup, they've never asked me a single leet code question, not once. Only time Don Nico was with Google and Amazon and another company that was like established, but they were not a FANG company.

They were just like a bigger not startup level company. And they mimicked the FANG interview process. But like an interview I had for a company, a great company actually, to be their founding engineer. It was all like, How would you solve this problem? And not only did I solve it, like I made it better.

Like they were like, oh, you can use like a fake database and just make it work we'll understand it. I actually spun up a database. I actually made it production ready. I actually got all this stuff done in the allocated amount of time and I blew them away. They were like, this is [00:33:00] remarkable.

This is phenomenal. But that's all because I did projects. That's only because I've done the repetition of spinning up like a local Postgres instance in a docker image that I could communicate like immediately, like hundreds and hundreds of times. I.

**Lane:** No, that makes perfect sense. I think that's a really good breakdown and it's, this is why I love having people like you on the pod because, most of my career I've worked at not like startup startups, but like smaller companies the company I spent most of the time at over the last.

Four or five years was a 600 person company, engineering team of 100. So like series D ish startup but like a startup. And at the end of the day, yeah, like most of our technical interviews, we might toss a quick coding challenge your way in terms of an algorithm or data structure. But at the end of the day, we were more interested in projects.

And I do think that, If I had to view my own bias, it's biased towards building projects, solving problems, demonstrating that things rather than lead code. But I do remember I had my interview at Google. I was like halfway into the interview process at Google when I got an [00:34:00] offer from a different company and didn't finish oh my God, they had me writing some crazy algorithm in a Google doc.

**Melkey:** Yeah. They still do that.

**Lane:** Oh God. It was terrible, man.

**Melkey:** a, it's still a Google Doc.

**Lane:** It's like the most painful thing I've ever done in my life.

**Melkey:** Yeah. Yeah. And you can't compile it. You don't know if your stuff's working or not.

**Lane:** Not only could you not compile it, they expect like the code to be sy tactically. Correct. At least my interview did. My interviewer did. It's like I'm getting like, I'm like, assuming because I'm in a Google doc that I should be writing pseudo code and he is this won't compile, you're like missing a bracket.

I'm like, oh no. That's how this is gonna be.

**Melkey:** Yeah, and that's why it's it's super important to very early on and like in that stream I talked about like before you jump into code because everyone you know is coding and all this stuff. You have to understand what your goal is because you can really screw yourself up if if you're doing a ton of projects for your nine to five, let's say, you, you were able to buy those 90 days and you're nine to five, you're just making projects.

But then you apply to Google like you're screwed. [00:35:00] And like the stuff that you just did is not applicable. But if you're granting lead code and you apply to Google, like that's way more applicable. And the flip side is if you spend your time doing projects, then you should probably catering to startups.

So you have to like really set your expectations early on just so you don't waste time. And I think a lot of people waste time. A lot of people waste. I'm gonna get into this industry.

**Lane:** Yeah, I couldn't agree more with that. And it is tricky because there's lots of different ways to do it and pretty much any engineer you talk to, like you go work on a team of 10 engineers and you talk to each one individually, they'll give you a very different story about how they got into engineering.

**Melkey:** Yeah, for sure.

**Lane:** it's crazy, like how many, different ways can work out in the end.

**Melkey:** Yeah, but it just, that's just like programming, right? There's not, there's never one solution to a problem. Everyone's gonna code a different one or implement a different data structure. Just like how people get to certain positions. Everyone's gonna have their own path. Everyone's gonna do it their own way.

It doesn't don't model yourself after someone else. Just do what works for you.

**Lane:** Fantastic. Dude. Thank you so much for coming on. This has been a ton of fun.

**Melkey:** For sure [00:36:00] man. Thanks for having me.

**Lane:** yeah, absolutely. Where can people find you online? What, where are you hanging out? What are you doing?

**Melkey:** Yeah, I think right now, so obviously Twitch is I think my most comfortable platform. So that's twitch.tv/milky. Used to be Milky Dev, but now it's just Milky. My YouTube is Milky Dev. I couldn't get the milky handle that was already taken that bastard. And then Twitter is twitter.twitter.com/dev.

Super active on all three of those.

**Lane:** fantastic. I'll link those down in the show notes below. Thanks again for coming on, man, and talk to you later.

**Melkey:** Yeah, this is some good backend banter. Let's do it again.