“I’m currently in Consulting/ Marketing/ Finance/ BizOps…, can I become a data scientist?”
I’m sure I’m not the only one asking this question.
When I first started thinking about the data science career, I tried to talk to a lot of data scientists. I wanted to know:
- Whether I’ll like this career
- Given my business background, how difficult or easy it is for me to become a data scientist
Scott Czepiel was among those who agreed to share their insights with me.
Scott’s intensive career of working with data as an analyst, researcher, statistician, analytics leader, and data scientist, has enabled him to look at the “Data” field as a whole, and give clear, practical, no-biased answers to my questions. (I’m sure he’s not one of those who think Data Scientist is the sexiest job of the 21th century, but I should ask.)
If you’re thinking of a data science career, I highly recommend you read on.
Q1: How did your data science career start? Did you have technical background and transition from there?
“After grad school I got started in market research. Along with marketing analytics, this is a natural fit for sociologists because it is fundamentally about understanding human behavior. It also involves more hands-on data analysis skills that aren’t as common in marketing proper. I moved to Amazon as a business intelligence analyst, before the term ‘data science’ had become popularized. Since moving to SF, I’ve been working mostly in the tech industry.
I’ve always positioned myself as having a blend of science, engineering, and art (yeah, I know it sounds pretentious!).
- ‘Science’ being statistics, experimentation, and the logic of data analysis;
- ‘Engineering’ being the ability to move data around, reshape it, and present it in reports, and
- the ‘Art’ being the ability to interpret data, combine it with domain knowledge and present findings to technical and non-technical audiences.
Whether it’s called analytics, business intelligence, or data science, I think these are the 3 cornerstones of the field.”
Q2: If you had to transition from a non-tech background, how did you do it? What did you have to learn and do in order to switch?
“All my programming skills are self-taught, but even after 20 years I still run into the ridiculous bias that I can’t possibly know how to write code because I don’t have a CS degree. At the same time, the programming that I’ve focused on is not the same that engineers specialize in. I pick up what I need in order to make data analysis easier, but there’s no need to become an expert in Python or Javascript to do that.
I actually think the tech skills are the easier part of the job. To really make a difference in an organization, to fulfill the hype or promise of what data science is supposed to be, is more about understanding the business, interpreting data in the right context to generate insights and make recommendations.
This is actually really hard to do and I know plenty of physics PhDs who know everything about algorithms but know nothing about how business works. Coming from a business background puts you in a very advantageous position because you are already familiar with how business functions, the process of decision making, product development process, business process, etc. All of these are much harder for someone with a pure tech background to pick up on their own.
The #1 tech skill to acquire first is SQL. It’s the foundation of accessing data from databases, it’s used everywhere, and getting more comfortable with it will unlock more job opportunities and set you apart from others.
After or along with SQL, try working with some visualization and reporting tools such as Tableau or Looker. At many companies this would be all you need. To go further, learn R. The best place to start is with RStudio and the excellent book, R for Data Science. This book will teach you the “tidyverse” which is a set of packages that has really modernized the R language and made it far more intuitive for data analysis than it used to be.
Data science is so broad that you don’t need all of these things to be successful. A good friend of mine who’s been VP of DS at several companies only knows Tableau. No SQL, no R or Python, just Tableau, but he’s also got a very keen insight so he’s successful because he can take Tableau reports and spin up an incredible narrative about the data.
At the end of the day, that’s what an organization needs from a data scientist: someone to look at data and understand what it means and tell the story about it.”
Q3: What do you like the most and the least about your data science job(s)?
“Most of my gripes about DS have more to do with the tech industry than about data science itself. However, the two are very much linked because a lot of the demand for data scientists is happening within tech. I’m trying to branch out to other industries (the health sector specifically), but especially here in the Bay area, jobs in tech are more common.
Like you also mentioned, I love to analyze data, and that’s always been the main draw for me. I also love the feeling of helping people understand something that they would otherwise not be able to discover because of having a different skill set.
Being a data scientist gives you time to be both an introvert and an extrovert.
I don’t know of any other job (besides HR) where you work with literally every department: engineering, product, sales, marketing, finance, customer service, IT, at some point you get to know all of them because they all need data and they all need someone to help them understand it so they can be more productive in their own domain.”
Q4: What do you think about the future of the data science job market?
“The demand for data science will continue to outpace supply for a long time. Yes, there are lots of people entering the field, but if you approach it from a “business first” perspective, there will always be a need for people to understand and interpret the data.
Data science is also very broad and means different things at different companies and to different people. You will encounter those who say data science is only about deep learning models, things like computer vision for self-driving cars, or genetic sequencing. That’s a very narrow view. Data science is about a lot more than that. Just be aware there’s a bias but mainly ignore those people. The bias is usually from genius coders with hard technical degrees who look down on everyone else. I think it’s much better to understand your strengths and weaknesses, let others do the jobs you can’t, and you do what you’re good at.
As Data Science matures, it appears to me to be specializing along 3 lines that I would roughly classify as business intelligence, product analytics, and machine learning engineering.
- BI is focused on data pipelines, reporting, dashboards, analysis tools. There’s lots of work with the business side to understand high-level requirements and translate those to engineers. Also, auditing data.
- Product analytics is a little bit of BI but focused more on driving product recommendations. It’s about becoming an expert on how customers use the product.
- The modeling side has lots of crossover with ML engineers. This is more about building data-driven products, where data scientists will do research, build prototypes and specify high-level what the flow of data should be, but then a dedicated engineering team will implement it in production. Even this is a very broad brush, there’s lots of cross-over and lots of companies where data science is a blend of all 3 or a different mix.
In summary, I feel very fortunate that I stumbled into this field before it became a thing, when I was just doing what I enjoyed. There’s lots of room to grow, and even better is finding your niche where you build up valuable domain expertise, which again I would argue is more useful in the long run than tech skills.”
If you want to learn more about Scott and his data journey, visit his blog “Variables and Observations” where he writes about work, data, and working_with_data.
Final thoughts
Thanks to Scott, I was able to gain confidence and made the decision to pursue a data science career.
My key takeaways are:
- Data science is broad and badly defined. Know your niche and develop your positioning in the market.
- Brush up your SQL skill!
- Data science = Science (the logics) + Engineering (the actions) + Art (the communication/applications). I think of it as cooking: you need to understand how to use heat, pressure, chemical reactions… to cook a dish, then you need the hard skills to actually chop, stir, fry… the food, and finally, a good presentation on a huge plate with a flower on top to convince people that this dish is delicious.
- That being said, now you just need to pick your focus, and work on the skills that need improvement. For me, it’s the Science and some of the Engineering parts.
Thanks Scott! And stay tuned for more interviews to come.
This article is also published on my Medium. Photo by Jeremy Bishop on Unsplash