This will be an opinion piece so if you have thoughts/disagreements please comment!
If you’re even tangentially in the data careers space or are in a predominantly Data role, you know how hot the Data Scientist role has been. Ever since that one Harvard Business Review article (you know the one), this space has been dominating ‘Top 10 Careers" and "Highest Paid Careers" rankings everywhere. Even now, you really can’t avoid being inundated with some article about this title; "Data Scientist job is dead", "Data Engineer vs Data Scientist", "Data Scientist is still the sexiest job", and so much more. Everyone’s got an opinion and if you’re like me – you’ve read it all.
I won’t talk about what you "should" do or be during your day job in this story; instead, I want to focus on what you do outside of your 9–5.
We all get into this career for a variety of reasons and I genuinely believe that variety enriches the field more than harms. Regardless of how the careers for this role will pan out, I believe that if you aim to being a Data Scientist outside of your day job then you will be forever desirable for your career and, more importantly, will live an incredibly meaningful life.
Before I make my case and lay out the framework of what I think this should look like, let’s get some ground rules set:
"I don’t want to do any ‘work’ outside of my 9–5"
This is NOT a proposal to spend more hours doing your day job. This is an extension of the ‘continuous learning’ mindset, but in a way that I believe will be more enjoyable and, over time, vastly more valuable to you.
If you’re not as interested in learning at least a little each day/week outside of your job then I’m making the assumption you wouldn’t even be reading this in the first place 🙂
"I don’t know coding and math or have very little of it"
Keep reading; technical skills are not a prerequisite for the life I’m going to propose. The only prerequisites are: curiosity and resilience.
"Who even are you to say what I do in my spare time?"
I’m surprised you didn’t ask this question first! This is what I have done/am doing in my career and I see how much value I gain from it. I want to share that knowledge with the hopes that others may also find a better path in the data field themselves. There’s a ton of ambiguity and frustrations when it comes to careers in this space. I’ve heard it range from a high bar for entry to a miserable life when the bar is achieved. I simply want to propose an alternative mindset/approach but your time is your most valuable asset – don’t let some random person on social media tell you how to spend it.
"…your time is your most valuable asset – don’t let some random person on social media tell you how to spend it"
Now that we got that out of the way, let’s start with time because I’m sure that’s the first thing on your mind.
How much time am I talking about when I say "spare time"?
The beauty of your spare time is that it’s strictly within your control. Obviously the more time you spend on things with focus, the better you’ll get at them. Regardless, you should first figure out how much time you really have available in your day. After your day job, what does your day look like? Do you stay up late or wake up early? Can you extract an hour or two a night or every other night to grow your skills? How much of your weekends can be purposed to self-enrichment?
I don’t have a concrete answer for this and that is reason #1 why I recommend this: your autonomy, by default, makes it a meaningful choice if you choose to make it. In other words, if you choose to spend time on learning something that you don’t have to do then it already has at least double the value than most other things you do in life.
Plan out your days/weeks, see how much time you can attribute to growing, commit and just do.
Now, do what exactly?
Does this mean doing Data Science work as freelance?
No. But it can if you want it to. What I mean is commit yourself to the practice of "making data useful" wherever you choose. If that means doing freelance work then great you can have a side hustle, but it very well can also not be for profit. How? Why? Let’s first start with "the practice of making data useful".
What do you mean by "making data useful"?
Which domain do you care about? Or better yet, which domain are you curious about? Are you interested how Tesla designs autonomous vehicles? Do you want to see how Spotify’s Discover Weekly algorithm works? Do you care about ACLU’s mission for an equitable and just future? Think about where your mind wanders for no reason; the things that make your eyes widen more than normal are going to be a key first step. And if you really don’t have that, pick the thing you have the most questions about. It’ll pay off just the same!
Now that you have a domain that you’re interested in/curious about, you have also inherited the problems that domain is facing today. Say you really care about eating green and reducing carbon emissions. You identify that the domain that interests you is environment/ecology, and abstracted it can be ‘social good’. Now that domain has it’s current problems it’s experiencing. Industry and academia are focusing on these problems today. A little bit of fairly straight-forward googling should be sufficient to find these problems.
Now I’m willing to bet that almost all of the problems you find for whatever domain you pick in today’s age will have some connection to data. It may be that researchers are collecting data to conduct an experiment to assess a question, or that some company is working with data to automate this problem away, or that data quality is quite low and it is causing more harm than good, or that they have quality data but not enough talent to help their cause. Almost each problem, or most problems, will have some tie to data.
Attempting to solve these [fundamentally data] problems is what I mean by "making data useful". Data Scientists are experts at this craft. They do it in their respective domains with a variety of skillsets and it can be an incredibly challenging because, as I’m sure you already noticed, these are hard problems.
How do I even begin to solve these problems, especially in my spare time?
You don’t need to solve the problem, I’m only proposing you give your best possible shot at solving whatever you can. The value here is not in you solving a massive impediment in your spare time; the value will be realized in your journey along the way.
Let’s say you care about ACLU’s mission and wanted to practice what I’m suggesting here. Assuming you’ve done your research on the problem and know it relatively well, I’d google if there are ACLU researchers or staff who are working with data. If you find them and find their publicly available email, RESPECTFULLY email them to inquire if you can pick their mind about this problem you’ve been researching. Of all the people you may email, maybe only a few will get back to you but that’s okay. Ask them what they’re working on and how they thought to try and solve this problem. Ask them what they’re needing assistance on and if you may contribute some portion of your spare time to assist for free (it likely won’t go favorably if you ask for money; that route is more for a traditional freelance route). Be very honest about who you are, what you can/can’t do, and what you’re looking for. Most likely, the ACLU will say something like "ehh I don’t know if I can just openly share this data". In fact, most people who you ask this to will say this. In response, I’d recommend asking if you can sign an NDA and then help. If not, then ask if there’s a way they can obfuscate the confidential pieces for you to assist on a smaller part of the larger project. If not, then at the minimum you’ve now made a contact with the ACLU and I guarantee they’ll remember someone who cold contacted them with a genuine passion for their mission to solve their problems. The solo route after this is check on Kaggle.com or Google Research’s Dataset repository to see if there’s an open dataset you can access (you can also do this route first if you chose). If no, then check if there’s an open API that they have made available for use. You may need to exercise some data engineering skills to link to it and obtain data, but still an incredibly valuable skill to learn.
My personal experience is people are really eager to get driven data practitioners help out with their problems. Assuming you’re able to get over the hurdle of getting data (signing NDAs, accepting a part-time internship, open API, etc.), this is 100% going to lead you to some amazing projects.
When you have the dataset, the sky is the limit. Use whatever Python and data skills you have (they don’t have to be impeccably perfect, usually just enough to understand how to begin to get what you need is good) and keep learning to make up the gap. Imposter syndrome is there for everyone in this field, but I promise it becomes easier to deal with if you’re working on something you care about.
What do I do if I don’t have any data, Python, statistics/math skills?
Another benefit of ‘being a Data Scientist in your spare time’ is that the broad definition works in your favor. There are an incredible amount of low-cost resources out there that will be a tremendous help. In fact, there are so many great ones, I even wrote a story about it: Journeying into Data Science.
The thing that is most valued in this space is application. How you apply the skills you learn and why you apply them there is some of the most significant criteria people in this space are focusing on. So the only change to the framework I’d recommend for very early beginners is after you feel mildly ready to start a project, look at Kaggle datasets and open APIs before reaching out to people. It’ll get you a little more comfortable with working on a project and practicing your skills before taking on something ‘real’.
This sounds great in theory, but does this actually work?
Here are two examples that have happened in my life:
The first is where I got to be an unpaid Research Assistant for computer vision project at University of Michigan. I was already attending the University for my Masters in Applied Data Science and during my off months I took this [amazing] Deep Learning specialization on Coursera. I scheduled my time outside of work and school to commit to this appropriately and was able to get pretty good with what I was working on. After some months, I got so good that I was finishing my coursework at a much faster pace. This left me with more time and I felt I was ready to try my hand at a real project. I started googling research projects at UM and saw one that perfectly aligned with my passion of art & photography. The research project was conducting computer vision on University of Michigan’s Museum of Art’s digitized art pieces to detect human representation over time based on race, gender, and age. Very difficult problem and I definitely didn’t have the perfect skills for it, but I knew how passionate I was about it so I applied. The interviews went well, but ultimately they chose a different student due to how many things I was already doing. I suggested if I would be able to let on for free. I would only be contributing my time and effort to assist and ask for no compensation; I just wanted to be a ‘Data Scientist in my spare time’. They had a hard time refusing that offer so they agreed and I got to work on this phenomenal project. I wrote scripts that got their data from their API, conducted extensive exploratory univariate and bivariate analysis, and also apply some cutting edge models to build out the MVP (story to come on this in the future). Eventually I had to leave the project before seeing it to the end due to scheduling constraints, but I was able to contribute quite meaningfully and saw my skills grow immensely in that time I was on it. I built some truly incredible relationships along the way and am quite excited to see this project come to life in an art exhibit at UMMA as it finishes! As a photographer, this project will always hold a special place in my heart and I’m so grateful I took the dive in the deep end to grow myself.
The second example is related to my passion for Sports Analytics. I’m quite a big sports fan, especially basketball. I knew I needed to work on at least some data project related to it but didn’t know what. After taking my Network Analysis course, I fell in love with the topic so much that I started perusing research papers in my spare time. I read a research paper about how the researchers analyzed Soccer Eurocup World Finals data through a network analysis perspective and my eyes instantly lit up. They were analyzing soccer players and games through data on how players passed the ball and I knew I wanted to try my hand at applying that for basketball. But how? I didn’t have basketball passing data! As suggested above, I googled NBA API and found exactly what I was looking for (story to come on this in the future as well). I read up on the documentation and through a bunch of articles to get familiar with the API. Then I wrote an ETL pipeline to read in the data and get it in the format that I wanted so I had all the passes a player made and received and by whom. Finally, I modeled that data as a network to assess it graphically and see if that research paper held up across sports. I ended up finding that paper didn’t hold up as well to basketball and then went on to create my own metric that I argued measured basketball player/team performance better. I already have ideas to expand on it even though just doing that much was quite the enriching learning experience and for a domain I was passionate about!
Okay.. so what?
These two projects are two of the projects I talk about the most even though others were more challenging or ‘technically savvy’. This isn’t because I think they’re the most flashy, but because I chose them when I didn’t need to. I sought them out and left my mark. They were about domains I was deeply passionate about and they were challenging enough that they made me far better at my craft after I moved on.
The ironic piece is that the projects I didn’t need to do in my spare time ended up being the biggest wins to progress my 9–5 life ahead as well. As I said before, the prerequisites for this life are curiosity and resilience. The more you can apply your skills and show you have those two in spades on top of your data skills, the more desirable you become. And the best part about curiosity and resilience is that they act as a positive feedback loop where they keep bringing wins into your life, but I’ll leave that for a separate story.
When you want to become a Data Scientist, either during your 9–5 or your spare time, you implicitly are already committing yourself to a life of consistent learning. You might as well make it learning about what you care about. We currently are living in an amazing age of data and open source where there is so much out there accessible to us, we just have to show enough persistence and respect for what we truly want.
Whether you’re a Data Analyst, Data Engineer, Machine Learning Engineer, Research Scientist, Data Scientist, or something completely different during your day job – aspire to make data useful in your spare time for the domains you care about and watch your life light up.
