I know, I know. The word "branding" makes me die a little inside too. But data science is a broad genre of work, and simultaneously serves as both a category of work and a job title, which makes finding your niche so important. So try not to go blind from rolling your eyes just yet; I'm going to put this in terms a Data Scientist can understand.
Take the below equation:
We know that in order to find y, we need the optimal combination of the other terms. Apply this to finding a model for your brand: let y be equal to "a career path in the data science I actually want". Can you picture what the terms describing y should be? In my mind, X is equal to skills, ß is equal to interests and ε is equal to expectations (more on that later). These terms work together to predict an outcome in data science, and we can interpret them similarly for developing our unique positioning as data professional. I call this method scientific soul searching.
Let's get searching
I'm going to walk you through developing your data science brand. Think hard about these features: the more honest you are with yourself about your skills, interests, and intentions², the more skilled and memorable you will be.
X - knowing your skills
Data science is a buffet. There are libraries and languages aplenty to keep your eyes glued to a terminal for years. Search job listings and see a broad distribution of skills: Python, R, SQL, viz, ETL, NLP, statistical analysis, deep learning, Hadoop, etc.
It would be tempting to try learning all of these skills to appear competitive, but this is the absolute wrong strategy. Each of these skills is associated with an entirely different skill set.
Holding all else equal, what are your strongest skills? Forget about Python, R, SQL and any of the other basic prerequisites for now³. Think of an interesting analysis you did, or the time you took a novel approach to dealing with sparse or missing features. What is at the foundation of each instance? Dig deep - use your soft skills to highlight your hard skills. Set yourself apart from the horde.
𝛽 - knowing your interests and what's important to you
Take a moment now to write down a handful of your interests, both professional and personal. Study what they have in common. Now go down the list, explain why you are interested in each.
Be true to yourself here. This step key. It's easy to get swept up in whats hot, and pretend to be into something you're not simply because there's opportunity this week. I say this because "people don't care what you do, they care why you do it."¹ If you are enthusiastic about a subject, and about how data science can impact that subject, will be apparent to both interviewers and the community at large.
You've got the X and 𝛽, now prove it with projects
This seems obvious, but I often receive messages from people who don't realize this to be true. You've got the brand, now you have to get to work! I'll share my brand story to demonstrate how integral public facing projects are to securing interesting work.
I had a handful of ideas for my Metis capstone project. I considered the formula I've presented to you: I wished to highlight my NLP skills but it wasn't immediately obvious how. I did know, though, that I was personally drawn to the Supreme Court, a topic I've been fascinated by for almost 20 years.
I pushed forward with an idea to topic model every Supreme Court case in history, excited to talk about it to anyone that would listen. It turned out that this enthusiasm was memorable: my project was passed around social media hundreds of times and this buzz resulted in a number of recruiter calls from tech firms like Google out of the blue. And the project has resulted in plenty of speaking opportunities as well.
I'm writing this post now as a pioneer for NLP in legal aid - building complex models that connect clients in need with the appropriate pro bono lawyer. This has been one of the most rewarding experiences I've had in my career, and it came to me because the founder of the startup saw my Supreme Court project online.
What can you do? So you've done a project with your X and 𝛽, now what? Tell me about it. Blog about it. Tweet about it. Connect with people directly. The more you write and talk about the subject, the more your name will become synonymous with your interests. The communication aspect is important because hard skills are nothing without being able to present them in a compelling way.
But wait, the dreaded Ɛ
It's easy to enter the market wide eyed and ready to impress. You've sorted out your angles and are confident in your unique voice. But you see rejection after rejection for the coveted title of Data Scientist. All of the pieces are there but your predictions are off.
Just as regressions without an error term, you've overfit the model. Take a step back; assess what roles are within your reach right now. Your ε is a good opportunity to consider where you want to get, but aren't quite there yet. Take the role that fits in your career (pro)gression while continuing to build your brand.
Parting words: always ask why
If you've made it this far, you should understand branding as more about knowing who you are and what work moves you than it is about choosing the right color scheme for the fall campaign. When you get into it, developing your brand as a Data Scientist is very fun, as you meet others with similar skills and interests.
- Simon Sinek's excellent talk on how great leaders inspire action.
- I didn't want to get too into intentions in this post, but don't be One of Those People who enters the industry solely for the money. If you are currently considering data science as a career, write down your reasons for pursuing data science and what you've liked and not liked about previous jobs. Be realistic about your current skills and abilities. Once you have this information, line up informational interviews to confirm with yourself. Hint: I have been known to occasionally accept informational interview requests.
- Are you an expert at any of the prerequisites? Can you describe a situation that proves that? Then absolutely highlight that skill! For example: you've been writing data processing modules in Python for 3 years and have contributed to sci-kit learn - your Python skill is your brand. The majority of Python programmers can't say that about themselves!