Daniel Bourke got hired as a junior machine-learning engineer at Max Kelsen after an internship. He's building five deep-learning models for a speech-to-text capstone project and experimenting with Edward, a Bayesian probabilistic programming library, to learn alternative inference methods.
Chapters
Yo, yo, yo! Welcome back to another week of learning intelligence. So glad you could join me here. And this week, this week's going to be really, really exciting because I've got some awesome news to share with you. But just one second, I want to show you what I've been working on. It's the VUI project for the Udacity Artificial Intelligence Nanodegree. And what's VUI? Well, voice user interface. It's the final capstone project of term 2, which I'm really excited So there we go, Voice Unit User Interface Capstone. And what I did is I went through the Jupyter Notebook and took notes on what I have to do for each step. So there you go, step 1, acoustic features, notes just in my own words of what's going on with the project. Some diagrams here I stole from it. There we go, important note and whatnot. But we have our first neural network training. Essentially what we have to do, or what I'm going to have to do is build 5 different neural networks from, from RNNs to RNN plus CNN, or recurrent neural network plus convolutional neural network. And the idea is to build an end-to-end deep learning pipeline to recognize voice and synthesize it into text. Or I believe that's the end goal of the project. But essentially, that's what it is. 5 deep learning models. Enough about that. We can, we can get more into that project later. I just got off the phone with the CEO of Max Kelsen, Nick, who— where Max Kelsen is a company I've been doing an internship at for the past few weeks, and I just essentially got offered a role. So my first day is tomorrow as a paid junior data scientist, junior ML engineer, junior deep learning specialist, something around that, but essentially an entry role starting 3 days a week, which, which I'm absolutely over the moon with. I was running around outside before on the phone. As you probably see, I like to walk around while I'm on the phone. But yeah, it's happening, guys, getting into the industry. And what's my goal? Essentially, it's to, of course, help them, but it's to gain valuable experience in what I've been studying. And 'cause nothing beats hands-on experience, right? So, as soon as I figure out more, there's some exciting projects going down on Max Kelson, and of course I'm gonna keep the videos going. So, don't— even though this is a sort of a little change in direction for what's happening for myself, expect these videos to keep going, right? They're not going to change. So, I'm going to still put out at least something once a week, whatever I'm learning, whatever's going, and yeah. So, I'll see you tomorrow after work, I guess. Have a good day, boys. Thanks, Mum. What are you working on today, Billsie? I have a mining company. Can't disclose the logistics about that. If I tell you, I have to kill you, right? Hello? It's lunchtime, and I realize it's pretty windy here, so I'm gonna try and make this quick. I want to share with you what I've been— share with you what I've been working on. So this morning And got in, did some reading, did some replying on emails. I mean, sorry, on LinkedIn direct messaging. If you want to get in contact with me, message me on LinkedIn. I've been really, really, really bullish on that platform lately. But what I've been working on in the data science role, I actually haven't got a title yet. I would consider it a little bit of data science, a little bit of machine learning, a little bit of this, a little bit of that. So once I find out the official title, I'll let you know, but I'm going to call myself a machine learning engineer or a data scientist for now. Now. Let's, let's just keep it at that. But 3 things, uh, for most of the work I've been doing this morning, it's been in JupyterLab. Now JupyterLab is like an online application where you can run Jupyter Notebooks, you can run Python code, you can do a whole bunch of different things. I'll leave a link in the description for all of this, by the way, so you can check it all out. So a lot of the work we've done is through the past couple of weeks in the internship as well has been through JupyterLab. And now What am I actually using in JupyterLab? Now it's all been in Python, but I'm using today, learning about, I've actually spent the whole morning learning about Edward, which is a Bayesian-based probabilistic programming library. So I mean, that's, that's a mouthful, but of course I'll link Edward as well in the description so you can read about it. But essentially what we're trying to do is we've run a deep neural network on some of the data, but now we're trying other methods to see if we can get some different insights. And the method we're trying today is Bayesian probabilistic programming. However, we've run into a bit of a roadblock. I'm going to go back after lunch and try to sort it out. Just trying to get our data in the right format to run through an Edward neural network. And then one last thing is, is how have I been learning about this stuff? Well, it's all been through YouTube. That's, that's how powerful the platform is. I watched a video from Siraj. I watched a video from one of the founders. I think his name is David Trang. I may be getting that wrong, but he's essentially a researcher for Google who helps develop the Edward framework. So that's my goal now, is to go back and try and get this model up and running. I believe I'm partnering with the CEO at some point today to discuss some formalities around the world, but I'm having so much fun so far, guys. It's, it's, it's, it's been really good. But yeah, if you want to ask me any more questions, leave a comment below or message me on LinkedIn. But I'll catch up with you once the day's over. Guess what? I found out what my official position title is. Machine learning specialist. But I'm gonna drop the specialist title and just say that it's machine learning engineer because I think that sounds a bit better to me, don't you? But it doesn't really matter. I'm having fun. It was an incredible day of learning, playing around with that new framework I told you about, Edward. I finished the day still with some roadblocks, but I left some notes for the other two engineers that that are on my team. There's a few of us. I haven't quite worked out the entire team dynamic yet because it's only 4 days or so in, but I'm excited. Tomorrow's a day off, but then after that we're going to get in and get back in and try to fix those errors that I come across. Oh, give me a second, I gotta get through these gates. Hey mate, how you doing? Thank you. So I have a total of 11 minutes to my train, so I'm gonna do some bit of writing, a bit of reading on the way home. It takes me about 35 minutes to get home. Um, yeah, I'm not entirely sure what's on the schedule for the rest of the week. I'm starting off at 3 days a week, Wednesday to Friday. As soon as I find out more, I'll let you know. Hey family, what's cracking? All your stuff's over here. Finished my first day, Mum. First day of internship. No, not internship, official day. Job! We made it. Woo! Yeah, what do you reckon, Mum? That's fantastic. We're home, team. What a day. Been learning a lot already, as you know, but I've got something cool to show you. I'll get it out of my bag, just give me one second. It's a bit crumpled up because it's been to and from. I hadn't really got a proper sleeve for it, but check it out! This is my part-time employee contract, all the official paperwork that means I'm an employee of Max Kelson. So thanks so much to the team there, really appreciate you guys hiring me. But there's one thing in here I want to show you, I kind of made sure it was added in at the contract. It's my favourite. I think you might know what it is. Fair work. There it is. This is the most important thing and the number one rule. Keep learning. Did you know that was coming? Did you know that was coming? I think that's an appropriate way to sign off the video, don't you think? Keep learning, guys. And P.S., if you want to ask me any questions about how I got a role or how I got an internship, I don't know the best answer of how you can get one as well, but I can share with you what's worked for me. So, leave your questions in the comments below. I'll answer them in a future live stream as well as I'll try and get to as many as I can or get to all of them practically via just typing. Um, but otherwise I'll record them all and we'll do a future livestream. But otherwise, peace out. And as always, keep going. Hello, hello.
