How I Figured Out What I Wanted To Do In Data Science & You Can Too

This decision is going to drive so much certainty around what you learn, what kinds of jobs/companies you target, how your career progresses, what types of projects and research you take on, etc. It’s one of those decisions that, if made early on, helps you avoid so much wasted time. The answer to many common questions about learning and career path can only be answered concisely by knowing what you want to do in the field.

I started in the same place most do. What are people willing to pay me to do? That’s what I want to work on. I was stuck in a bad work environment in a business that was on a deep downswing. I was looking for a way out and more than willing to take whatever came my way. Many in school or the work world are in the same boat. Employment and earnings are a powerful motivator but not a long-term motivator.

While that may drive your first job or other short-term decisions, that’s not a good way to direct your career. My first few consulting gigs were in the manufacturing and marketing space. I learned around very specific applications in two very narrow fields. After a couple of years, I realized that many of the broader machine learning and data science approaches were passing me by.

I took a data science certification as a catchup and refresher on other areas of the field. It helped me understand the broader implications/approaches of the natural language and computer vision work I was doing. Those two areas became my focus.

I’ve been able to apply both to a broad range of industries; retail, finance, technology, and recruiting. Focusing on two areas has also helped me keep up with the changes. There’s so much to learn and it’s changing so rapidly. Without focus, I’d be a generalist with far less value or capability.

I work with a lot of supervised and unsupervised ML, but my focus is on reinforcement learning. I see a great deal of potential for RL across a broader set of applications. I’ve chosen the applied side of data science. I consume research and build for production. I have a lengthy background in software development and management so it’s a natural fit for my mindset.

I’m aiming the future of my career at the strategy side of data science. About three years ago, I had a client ask me to build out a data science team for them. It’s one of the most enjoyable projects I’ve worked on and I’ve built out three more teams since then. I’ve been an advisor to businesses more than a practicing data scientist over the last three years.

Around the same time, I had another opportunity present itself based on something I said while I was on a panel. After a ML company founder finished explaining their tech, a hedge fund manager in the audience said, “When I hear stuff like that, my crap detector goes off. That sounds like every other product in the space with the words ‘machine learning’ sprinkled in. How are we supposed to tell real from spin?”

After a few garbage answers from the panel, I cut in with a five-minute response on due diligence for machine learning products. I have a tendency to be a bit too honest about our field, so I don’t often get invited to conferences. If you don’t want to hear my thoughts, don’t invite me. If you’re going to put on hype-based content, don’t expect me to be an active participant. Like I said, not a lot of conferences want me calling out the overabundance of BS that goes on there.

The hedge fund manager grabbed me after the panel was over and asked if I could talk with his team the next day. Sine then, I’ve discovered two audiences who are looking for brutal honesty because they don’t get it very often; senior leaders and investment managers. I’m brought in four or five times a year to speak to groups about the realities of machine learning.

I’ve been called a thought leader and influencer a lot over the last four years. I don’t really know how that happened. Cat videos get more engagement than my tweets. I’m someone who’s on the ground in the field and saying what everyone else is thinking. More and more, I’m seeing others talk openly about the same things. Am I really leading anything? I’m not sure if I’m being realistic or suffering from imposter syndrome.

Lessons From A Meandering Path

As you can hear, my path was not direct. That’s what I want to help you avoid. Most of the lessons I learned were by chance and fortunate accident. Obviously, that’s less than ideal.

Your first job will be more a matter of necessity than choice. However, that’s not how your learning path should be. Choose to learn concepts that bridge different industries. I’ve chosen computer vision and natural language but choose the one that interests you most. You can also learn the broad set of ML used in a specific industry. For example, you could learn ML with specific focus on how they’re applied to manufacturing or autonomous vehicles. The choice boils down to either a field of machine learning or a specific industry. Learn like a generalist until you find the one that interests you most, then specialize.

Choose applied or research data science. This is a much simpler choice. Do you enjoy the research, publishing, and theory of machine learning? Do you enjoy building for production? Try both and focus on one or the other.

Look forward a bit. What trends do you see as promising or interesting? What models that are being fleshed out now do you think have the most potential? Where do you see ML adding the most value? Learn and grow in directions you think will support your career over the long term.

Be you because what makes you different is what makes you valuable. Not everyone will see your value. That’s no reason to try to change your value proposition in an attempt to please everyone. You’ll be different from others which can be scary when you’re starting out. Gravitate towards those who see a need for what you provide. Advertise yourself to the niches you find. There is more value in a hard to find individual in even the smallest niche, than a conformist. The conformist tries to be just like everyone else making them very easy to replace.

Get used to saying what’s on your mind. The longer you’re in this field, the more valuable those thoughts become. If you don’t start talking early, you’ll never find your voice, and no one will ever benefit from it. Your voice is how you give back and your best vehicle to showcase your value.

There are few data scientist and fewer good ones. If you focus your learning early, you’re going to be one of the good ones. Find your voice and you’re going to be in the very top of the field. It’s a weird place. Half the crowd there are talking heads. You’ll be a little miffed to be lumped in with them. The other half are people you’ve learned from and you’ll have a healthy skepticism of your place amongst them. Every time I’m on a list with Kirk Bourne, Gregory Piatetsky-Shapiro, or Carla Gentry I just laugh. Be real, there’s no way.

On the other end of the scale is full blown imposter syndrome. It’s crippling when you don’t believe in yourself. A structured journey goes a long way towards overcoming imposter syndrome. If you’ve built a path to the top, it’s a lot easier to accept when you arrive there. It’s also easier to stay grounded. Both are important because all that matters is your next project. A structured journey makes that focus possible. Otherwise you’re always being pulled in some direction by external forces.

0 replies

Leave a Reply

Want to join the discussion?
Feel free to contribute!

Leave a Reply

Your email address will not be published. Required fields are marked *