My two cents on what makes a good data scientist nowadays?
My parents recently asked me what I do for living again. In the past, it was not that easy to explain as being in data science was not a well-defined field. Even when I started my first job at Accenture, I barely knew what I really do. Of course there were some definitions like:
Or the famous “Data Science Venn Diagram” by Drew Conway.
So at this time, I had some kind of clue what data science is but from my perspective everyone in my field including me was still figuring out what data science is and how it can bring value. At least, this is what the majority of people did. Of course, some folks at Google or Amazon understood it right away of what I would call modern data science nowadays but others were just experimenting with it. For example, while at Accenture most of the clients I worked with “were looking to make use of their big, very big data”—whatever this means. I’m sure many of you have heard this line before especially if you are working in this area. This goal was so vague that most of the project that I did was kind of proof of concepts that ended up on some powerpoint presentations.
Now after some years in the job, I fairly have a better understanding of my job which is simply to create data products that are smart and data-driven— at least, this is what I tell my parents nowadays and they seem to understand this. One question that would pop out now is what qualities are now important for a good data scientist to stand out these days?
Good data scientists are also good product managers
Many well known digital products are directly powered by data science and machine learning. For example,
- Google uses machine learning in its search algorithms to improve their search quality
- Spotify, Netflix or Amazon use recommendation engines to recommend new music, movies or products
- Facebook uses computer vision techniques to automatically tag your friends in their photos even if it can’t see their face
- Uber uses machine learning to create self-driving cars and many more things.
As you can see from those examples, disruptive technologies make good use of data and they are part of an overall product.
In these days, it’s not enough to just toy around with data, creating models that end up in a powerpoint presentation but they need to be part of a product that is shipped! That’s why good data scientists need to be good product managers as well. Good data scientists need to understand how their work end up in a product and how it will be consumed. They also need to be able to create a product even if there is no data. I’ve experienced data scientists who told me that they only can do data science if there is data. Surely, this makes sense but not every product starts with a lot of data at the beginning. Or there is also no data especially if you are still figuring out what that product is. Thinking in terms of what will soon be possible with data is a very important skill. Trey Causey also recently published an article about this topic “Rise of the Data Product Manager”. You should check out his article. I also worked on this topic for a while. Last year I wrote an article which is called “API First for Data Science” which basically argues that data science models should be wrapped as an API as early as possible so that they can be seamlessly used in a product.
Good data scientists are also good soccer players
In soccer, you have eleven players. One player plays goalkeeper while the other ten plays as forwards, midfielders, or defenders. Although there are players that stand out like Cristiano Ronaldo or Lionel Messi, they are nothing without their team. For example, Messi is doing extremely well with his club FC Barcelona but when it comes to the national team where everything falls on his shoulder, his team usually does not perform well during a competition.
This also applies to creating amazing products. Although there are those super amazing unicorns aka data scientists out there, they are nothing without their peers like product managers, designers, software engineers, data engineers, functional experts or whatever roles are needed to create a successful product.
For example, let’s take Spotify as a great use case for a collaboration between design+data science (surely, other roles like product managers, software engineers are involved as well but design and data science are more obvious in this case). Here is an image of their product across multiple devices:
I just love their slick, dark, clean design which looks perfect on every device that I use. But the main reason why I use Spotify is because of their “Discover Weekly” feature where they compile a weekly personalized playlist based on what I’ve listened to in the past. And I must confess it’s working pretty well.
Wait — What about other qualities like data scientists are also good hackers, good statisticians, good communicators, good problem solvers etc…
As a matter of fact, everyone of us can agree that a good data scientist nowadays needs to have good programming, statistics, math, communication and problem-solver skills. There are many more skills but I intentionally didn’t talk about those ones because there are so many articles out there that discuss this. I only mentioned two areas that I see where data science will evolve. For example, we see a lot of data product manager roles coming up because many companies realize that many successful products are data-driven.
Moreover, there is also more interdisciplinary collaboration coming up in the data science space. Google started a new project that is called “human-centered machine learning” which basically should create some guidelines for designers working with AI. There are more collaboration coming up. I’m looking forward to this development.
Some More Readings
I’m sure many others will have a different opinion on what qualities make a good data scientist nowadays. I’m quite happy to work in a field that is constantly changing every day. So I would be happy to get other’s opinions on what qualities are/will be important?
Give me a ❤️ if you liked this post:) Follow me here on Medium Dat Tran or on twitter @datitran to stay up-to-date with my work.
