This article was adapted from a podcast. Listening/viewing options, as well as a full transcript, available here.
Last Friday, I answered audience members’ questions on getting started with learning machine learning. This week, I’m building on that by focusing later in the professional journey and handling questions on how to become a leader in machine learning.
Let’s kick things off with a question that the data scientist Mitchell Reynolds posed to me in a LinkedIn post. For a bit of context, Mitchell works out of San Francisco for Hint, a popular flavored water brand in the US.
Mitchell asked me: “If you could travel back in time, what advice would you tell yourself when transitioning from Data Scientist into a leadership role?”
As sad as it is, the more management responsibility you take on, the more time you spend in meetings and the less time you get to spend having fun writing code. I harbored a lot of guilt about that for several years because I felt internal pressure to be as on top of machine learning theory and production model deployments as the data scientists and software engineers on my team.
It sounds silly now that I felt that way because, of course, it’s impossible. The more people you hire, the higher their caliber, and the more independence you give them to try things out on their own (which you definitely should!), the less you’re going to be able to get in the weeds with them on every single thing they’re doing.
So my advice to me when transitioning to a data science leadership role would be to accept that, despite your best efforts to make time around management responsibilities, you will only get to spend a tenth as much time as you'd like to training models yourself... and that is ok. Indeed, it’s exactly what your team and your company needs from you.
Mitchell also asked about what my workweek is like as a Chief Data Scientist, but I have such a long answer for that that I’m saving it for a future episode at some point.
Finally, noting that he has a Bachelor’s degree in Economics, Mitchell also asked, “What is your and your peers' background for Data Science leadership? Do they have a PhD or a masters? Did they study computer science? Math? Economics?”
First I’ll tell you my academic background in one sentence. During my neuroscience PhD, I taught myself machine learning in order to be able to find patterns and causal relationships in large medical sciences datasets like brain imaging and whole-genome data. That said, despite me having a PhD in a quantitative discipline, I don't think any particular background matters. At all.
One of the best data scientists I've ever had the pleasure of working with has a bachelor of arts in the humanities. He could easily go toe to toe with, probably outdo, the vast majority of machine learning PhDs on a given machine learning problem.
There are two things that matter to be a data science leader: First is that you can continuously teach yourself and those around you; the space moves quickly and is extremely broad. Thankfully, you can learn whatever you need to know from free or inexpensive resources like YouTube, Udemy, SuperDataScience.com, Udacity, arxiv, the Khan Academy, and the O'Reilly learning platform.
Second of what matters is that you're willing to constantly apply what you've learned, that you're not afraid to make mistakes. Never stop learning and never stop applying and eventually, at some non-specific point, you’ll find that you’re a leader in the data science field no matter what your formal academic background was.
That’s all for this week! I hope you enjoyed the format I used for this and the preceding two articles in this series, wherein I answered audience questions on a specific theme each week: Two weeks ago, it was Future-Proofing your Data Science Career. Last Friday, it was Getting Started in Machine Learning. And, of course, today it was How to be a Data Science Leader.
Let me know if you liked this Q&A format! If you did, feel free to ask me your own data science or machine learning questions–or anything at all really–by tagging me in a post on LinkedIn or on Twitter @JonKrohnLearns and I’ll aim to answer your questions via social media or an upcoming SuperDataScience episode!