MLA 001 Degrees, Certificates, and Machine Learning Careers

Welcome to the first episode of the exclusive podcast Machine Learning. Applied a companion podcast to machine learning guide, and thank you for becoming a patron to unlock this series. I very much appreciate it. This series is gonna be a much more applied approach to machine learning. We're gonna talk about degrees and certificates, interviewing tactics, tech tips and tricks, stuff around Pandas, NumPy, TensorFlow, maybe any sort of hot news that's going on. The machine learning community, I. All that stuff that fits into a traditional style podcast where the Machine Learning Guide podcast is a little bit unusual in that they're very long one hour episodes that are educational and sequential. Machine learning guide is more what you'd get out of iTunes, you or the great courses machine learning applied. This podcast is gonna be like your traditional podcast. In this episode, we're gonna do a throwback to a machine learning guide podcast on certificates and degrees where we talked about Coursera specializations, Udacity and ano degrees, versus a bachelor's, master's, or PhD. And the reason I'm touching on that episode again is that if you recall in that episode, my interviewing experience in the machine learning industry was a little bit limited. And I said to take my recommendations there with a grain of salt. And since then I've gotten many, many emails people asking if my recommendations there still stack up if my recommendations in that podcast still hold. Especially the big question people keep asking is, should I get a master's? Should I do the O-M-S-C-S? Georgia Tech Online master's degree. And just a spoiler alert for this episode, the answer is yes, you should get a Master's. This is from personal experience for the last year since I released that episode, interviewing at many, many companies. So let's take it from the beginning. In that episode, I was talking about a few different types of people. Some people are just curious about how machine learning works from a high level perspective. My podcast is Akay for you. If you're a manager and you want to know what makes your employees tick or what kind of people to hire, all that stuff, all you need is the Machine Learning Guide podcast. Take it a step up and you're somebody who self teaches, really wants to learn machine learning, but wants to do it on your own. Diamond Time, an auto didact. We call that somebody who self teaches Well for you, I say listen to my podcast and follow the resources. The resources in each of the MLG podcasts. Show notes, those will guide you down the path of learning the details of all the content I discuss. You could do that, follow my podcast and resources, or you can do a Coursera specialization or Udacity nano degree now of the two. Coursera versus Udacity. I recommend Udacity, the Udacity nano degree. That's the most commonly discussed and recommended of the MOOCs certificates in machine learning. I see online and the closest of them all to being respected in industry. The closest. Now I say that, I mean that Udacity nano degrees are still not respected in industry. They won't get you a job. They are not an accredited degree. This I know from personal experience now from many conversations over the last year with hiring managers and people in industry and reading conversations online with recruiters and blah, blah, blah, Udacity Nanodegrees are still not currently respected certificates to land you a job in industry. They're fantastic certificates. They teach you a lot. They show tenacity. They show independence and self-teach. And importantly and importantly, what you'll come away from a Udacity nano degree with is a portfolio of code that you can use to show perspective employers during an interview, which is vital. Vital. You want a portfolio. You need a portfolio before you start applying to jobs. No matter what your degree is, no matter what your certificate is, you need some code. You need one, two, maybe three GitHub projects where you can walk your perspective employer through during an interview showing them how you handled data munging and cleanup. With Pandas and NumPy, exploratory data analysis, and then the modeling process, whether you're using Psychic Learn or TensorFlow, all that stuff, you need a portfolio before you start applying, and Udacity will give you that portfolio, but that's not all the way enough, unfortunately. I truly believe that Udacity is the future of education. I think that MOOCs are the future of education, and in particular, Udacity is showing the most promise and it has huge companies behind it and very high quality content, but they're certificates that you come away with. The Udacity nano degrees are still just not accepted widely enough in industry for you to confidently graduate that program thinking you're gonna get a job, Udacity is the future, but circa May, 2018. It is not yet the present. And so that brings us to our next topic of bachelor's versus masters. And here I'm just gonna tell you the answer is master's. You really want a master's degree to start applying to jobs in the machine learning world. Now, some anecdote, I used to be a web developer, react js, node js in Postgres, MongoDB React native mobile apps. Really that traditional direction. You throw your computer science grad, I have a bachelor's degree, not a master's degree. With that bachelor's degree, I really had no problem finding employment as a web developer. In fact, that bachelor's degree was really superfluous and I had a lot of conversations with friends that I even resented getting my degree because I used the subject material so seldom in industry, I mean a lot of these really deep data structures, algorithms type programming is. Inbuilt in the frameworks and libraries you're using, and furthermore, that my education was just not a factor in finding employment. A lot of my web developer colleagues were either graduates from other programs, political science, English, something that's a little difficult to find work in, or they didn't have a degree. They graduated from some bootcamp or did Code Academy or Code School or one of these things, they built up a solid portfolio and then they were able to land a job. In fact, I think that's a bit of a traditional story in web development and mobile app development, but in machine learning, while having a portfolio is essential before you start applying to jobs, it's not enough. When they say on their job descriptions, master's or PhD preferred, that's kind of a hard requirement. At least it's a, it's a harder than soft requirement. When they say Bachelor's preferred on a web developer job description, bachelor's degree tends to be a very soft requirement. In fact, it rarely came up with a lot of colleagues that I saw landing jobs in web velman. But in machine learning, a master's degree is a very, very harder than soft requirement. Now, I say it's not exactly a hard requirement because I have found jobs with my piddly bachelor's degree in machine learning, and I also have friends who have done so. But I want to tell you that it has been an uphill battle. It was a difficult process. It was. It's been difficult proving to my employers that I'm worth my Salt Sands, my Masters, and if I'm gonna be completely honest with you, the only reason I've found employment in machine learning is because of this podcast. If you recall from MLG, I mentioned in some episodes that I was looking for work and I had a lot of people contact me and it was fantastic. I'm glad I did that because that's how I've found gigs. Any jobs that I've directly applied to myself on LinkedIn or Indeed locally through meetups. I haven't even gotten an interview. I have not gotten an interview. And after talking with recruiters who really want to get you a job because they take a cut and asking them why I've had such poor experience, they told me it's because of my degree that I lack a master's. You don't need a master's to get into the industry, but. It is gonna be a battle without it. So I highly recommend to you to get your master's if you already have your bachelor's. It's only another year, give or take, if you do the OMS Cs, Georgia Tech Online Master's degree, which is what most people are doing when they're getting their master's in machine learning. These days it's about $7,000. It's about a year. Cheap and quick if you already have your, as long as you already have your bachelor's. So that's what I recommend for you. If you want a job in machine learning and you want to stand on equal footing with your competition and not have to push this boulder uphill, then get a Master's and just get it over with O-M-S-C-S, Georgia Tech online degree. And on a personal note, I'm gonna get that degree. I'm gonna get that degree as soon as my life kind of slows down a little bit, I have some time on my hands. I'm going to get that master's degree that just tells you how much I believe that that's essential to your perspective employment. If you're gonna try to crack in the industry without that master's degree, you're really gonna have to differentiate yourself. Make yourself stand out, make yourself special, build a powerful or popular online project that you can really just blow your interviewer away with GitHub code or number of users signed up to your service or something like this, or heck make a podcast. But even after saying all that, I mean, come on. Just get the master's degree. Alright, so that brings us to PhD. Do you want to get a PhD? That's up to you. PhD of course, is much more research centric. You're likely gonna be working in much more interesting projects, maybe for much more interesting companies like Google and OpenAI. I. Doing real cutting edge research in artificial intelligence. PhDs make a lot of money. There's been a lot of posts recently on Hacker News and Reddit of salaries in Silicon Valley for machine learning PhDs being in the $300,000 range, and this being somewhat of an average, not an extreme. I mean, that's a lot of money, but there's a major trade off, which is time. The amount of time you sink into getting your PhD is enormous, many, many years, and many, many hours per week during those years. So what I recommend is get your Bachelor's. If you don't already have it, then go onto your master's. Finish that up, and by the time you're done with your master's, you'll probably know personally for you whether research is the way you want to go. If not, you jump into industry with that master's. If so, then you just continue on where you left off and go on towards research. I personally will not be going that route. I'll stop at my master's. That's a wrap on certificates and degrees.