Journey of our Past Fellows

How I found my Career Path with Fellowship.ai

I must have been about 4 years out of college when I had the thought “data scientist, that sounds like a good idea.”  I’d been through a number of odd jobs that weren’t going to lead to anything (or, more accurately, I wasn’t going to make into anything).  I’d worked in a hospital, restaurant, chemical plant, multiple research labs, done a few odd catering gigs. Traveled until I ran out of money.  Lived off the grace and kindness of my friends’ living room floor. I’d gotten laid off, fired, and quit.  And really, I had absolutely no idea what I wanted to do.  

When I looked at what my friends from college were doing, particularly those who had studied physics as I had, I realized that the successful ones had generally followed one of two paths: either continued into academia in masters or PhD programs (for which I had neither the grades nor the inclination), or had learned to program and gotten jobs in tech.  And when I talked to my friends in tech, most of what I heard boiled down to two options: either web dev or data science (I know now that’s not quite true, but it was certainly my impression at the time).  Creating websites seemed boring and data science sounded fancy.  Science, I know that word, I studied that (little did I know data science is distinctly not science).  I’ve done some math; I understand calculus and linear algebra.  Machine learning, I like the sound of that. I know what regression is and what it means to minimize a loss function, I’ve already got a leg up. I even enjoy thinking about these things from time to time. And wow, there are so many jobs, everyone wants to hire one of these guys. I can barely write “hello world,” but we’ll figure that out later.  And there it was - data scientist, that sounds like a good idea.

So how does one begin to ‘learn data science’?  Well, I’ve got some friends that can code, I’ll  get some guidance from them. They said I should work through every problem in Learn Python the Hard Way, and that there was absolutely no way I’ll work through every problem on Project Euler but I better try none-the-less.  Oh, there is this new thing called MOOCs, let's see what those are about - I did the Johns Hopkins Coursera Data Science Specialization. I think I need a little more in-person guidance, it’s time to shell out - I enrolled in the General Assembly part time Data Science course at night after work.  

And then it was time to take the plunge.  I was accepted into startup.ml (since rebranded to fellowship.ai) - it’s brand new, I’ll be a first cohort guinea pig. It's a risk, but it’s now or never.  I quit my job, I enrolled in the 4 month fellowship, I moved back on my friends’ floor, I lied to them that I knew when I’d be moving out again, and I dove in.

So what did I learn in Fellowship.ai?  Well, a lot about data science and machine learning, sure.  When you do something full time, you’re going to learn a lot about it.  But mostly what I learned was how things work. That if something needs to happen, you need to do it.  If you don’t know how to do it, you need to figure it out - whether that’s through asking or on your own. That there isn’t always someone looking over your shoulder, but there’s always someone who will know when things don't work. They say that school is about learning how to think, but I was learning something I’d never learned before: how to do.  How to make happen. When to ask questions and when to figure it out as I go along.  Learning the confidence that I could learn whatever was needed, when it was needed.  And that that’s the way it happened in real life - your boss doesn’t have an answer key for the job, nobody knows exactly the ‘right’ answer.  Which is not to say I couldn’t have learned this in any of my former positions, but I hadn’t learned it.  That was the time things lined up for me to absorb these lessons.

I’d love to say that post fellowship I had my pick of jobs, but that’s not the way it went.  I interviewed around at a few places, but never got to the point of an offer.  I wanted to move to NY and I used my fellowship connections to score an interview I never could have gotten on my resume alone.  I didn’t wow them, and I didn’t receive an offer. But I’d learned how to make happen, so I asked about an internship, they agreed, and I moved to NY.  3 months later it became a full-time position, and I’ve worked there ever since - coming on six (!) years.

So that’s it?  Got the industry job, settled in, and I’m done?  Well no, not quite.  I’m still learning, still evolving, and still progressing. Since I’ve started my job I completed a Master’s degree in CS to fill out the gaps in my knowledge there.

I distinctly remember one of the speakers who visited the fellowship talking about his feeling that actual progress is made from machine learning by either the ML practitioners learning the domains to apply it to, or the domain experts learning ML.  Data Science is a tool, and I’ve learned how to use that tool.  But just because someone can swing a hammer doesn’t mean they know how to build a house.