Many of my followers ask me how difficult it is to get a job in the data science field. They also want to know what they should study or what path they should take to get involved in data science.
Unfortunately, the answers to these questions are probably not the ones everyone would like to hear. Getting into data science can be pretty difficult, and you have to work hard to get a start. Of course, data science isn’t unique in this regard. Lots of lucrative and interesting careers are hard to enter, and data science does offer a lot of benefits to its practitioners.
I’ve written here before about all the training that goes into making a data scientist. Not surprisingly for a field that requires such specialized knowledge, breaking into a data science career is pretty involved. Before beginning, you have to devote time to learning about the field and understanding algorithms. Later, you need to constantly upgrade your skills as the market progresses while still staying up to speed on old, conventional techniques. You also have to work on understanding the problems you’ll be solving for businesses and develop the acumen to frame business problems as data science problems. After all, since there are no fixed algorithms, every problem has its own unique solution. On top of all this, you’ll need to deal with searching for a job and preparing for interviews.
All this work is almost impossible for people who are unwilling to devote the necessary time. In thinking about myself, I know that I get bored quickly if I am not learning new things. I like data science as it gives me that opportunity. So first of all, you should ask if that’s true for yourself too.
If it is, and you’re interested in solving new problems almost every day, then you will love data science as a field to make your career in. And here are some tips for you brave ones.
1. Take Small Steps
According to an old proverb, it is better to take many small steps in the right direction than to make a giant leap forward only to stumble backward.
I once posted this proverb on a social media site and one of the commenters called it the gradient descent rule. I find that name fits pretty well. As you might not want to take an overly grand step in gradient descent optimization, you surely don’t want to do so when you start learning data science. This rule is even more true if you’re coming to data science from a different stream (fields with a lot of cognate skills like computer science or statistics excepted) or if you want to make a lateral switch. I’ve seen many people bite off more than they could chew when they start and then inevitably get so frustrated and disheartened that they abandon the field altogether. For example, I’ve watched as people tried to start by learning deep learning right away, only to inevitably get lost. A small step in the right direction would be learning some basic linear algebra or working to understand a basic model like linear regression first.
Furthermore, I would advise against targeting jobs at big companies like Amazon or Google as you start out. I offer this advice not to discourage you, but more as a matter of practical thinking. Having observed the interview processes of big companies like these, I can assure you that it’s pretty rare, if not impossible, to get these jobs without some experience. Your time is probably going to be far more effectively spent targeting roles at smaller companies.
After all, my advice doesn’t mean that there is a shortage of opportunities elsewhere. You can easily get in with a startup if you know your stuff. Ultimately, starting your career at a smaller company will also be of great benefit to your growth as a data scientist.
In fact, that’s how I started out myself. Experience at a startup helped my development as a data scientist immensely. I got to work on different problems end to end, took ownership of most of them, and talked to business stakeholders from day one. The startup also provided me with ample opportunity to invest time in experimenting, trying out new things, and reading up on new technology. That experience helped mold me into the data scientist that I am today.
2. Learning Is Paramount
“Nothing will work unless you do.” —Maya Angelou
I made it my goal to move into the data science space in 2013, and it has taken me a lot of failures and a lot of effort to shift jobs. I started my career as a business analyst in 2010. For three years, I did a lot of grunt reporting work including Excel/SQL and Spotfire. I only got a chance to try my hands on a data science problem by accident at the insistence of my manager. Working on that problem ignited my excitement to learn more about data science. From 2013 onwards, I have spent whatever free time I could find to study new technologies and grow as a data scientist. As a result, I’ve realized that the studying will never be over. Because the field is constantly evolving, there are always going to be some things I won’t know. Fortunately, the necessity of lifelong learning works for me since I love to learn new things.
If that’s also true for you, here is the sequence of courses that I took to learn data science. An aspiring person willing to put in the hours of effort can choose to become a self-trained data scientist through a similar course of study. The training involves a robust mathematical grounding along with a lot of practical project work that involves applying the theory learned in these courses to various practical scenarios.
I hope that you don’t lose hope after seeing the long list. You have to start with one or two courses. The rest will follow with time. Just remember that time is a luxury you can afford because it’s more important to be thorough than fast.
3. Work on Your Portfolio
Having a grasp of the theory is excellent, but you really don’t add value for a company as a data scientist if you can’t write code. So find ways to work on your coding ability as much as you can. Create stuff. Try out new pet projects. Go to Kaggle for inspiration. Participate in the discussion forums. But don’t stop there. Think creatively. Build your GitHub profile. Try to solve different problems.
For example, in my own starting phase, I created a simple graph visualization to discover interesting posts in the data science subreddit using d3.js and deployed it using Flask and Heroku. In addition to working on the usual data science problems, I created a blackjack simulator, and I also implemented a code-breaking solution using MCMC. These were all fun projects that also allowed me to experiment with theoretical concepts. I also took part in various Kaggle competitions, and though I don’t have much of a leaderboard rank to show for it, I ended up learning a lot.
In addition to serving as invaluable, hands-on training , you can also marshall this experience in your job search. Below, as an example, is the section about this work that I put in my resume when I was in the initial phases of my job search. This allowed me to showcase for a prospective employer that I was interested in and gaining experience relevant to data science.
4. Blog a Little
Honestly, I love to write, and although this step is not a definite requirement for becoming a data scientist, it can help a lot when you’re starting your career. I noticed that I understood data science concepts much better when I explained them. Blogging is a perfect tool for accomplishing this.
When you blog, you not only end up creating high-quality content for others to learn from, but you can also document your own learning, understand concepts better by explaining them, and maybe gain some extra recognition. What else would you want?
Also, data science is pretty vast, and I tend to forget whatever I learned some time ago. Blogging solves this problem too. I started my blog in 2013 and updated it with whatever I learned. Thus, I ended up documenting everything. I still consult my own blogs whenever I feel stuck on some problem. Furthermore, blogging has also helped me with my communication skills as it forced me to explain difficult concepts in simpler words.
If you don’t like the idea of blogging, you can achieve something similar by taking notes. As I said, blogging is a personal preference. And if you’re interested and want to know how I started writing on Medium, here is my story. Also, my blog has helped me to create a decent GitHub profile and gain more visibility. In fact, here is a popular GitHub repository of mine where I keep all codes for my data science posts.
5. Don’t Be Too Selective
I get a lot of queries from people in my LinkedIn network that go something like this: someone has received an offer from an analytics company, and they want to know if, by taking it, they’re saying goodbye to a career in data science. The main reason for their concern is because they’re afraid of accepting a role whose title doesn’t officially say the words “Data Scientist.” A related fear is that people often think that their first job title decides their career expertise forever. People are worried that future employers may pigeonhole them, considering them for only that specific area of expertise and nothing else.
What I inevitably tell people who ask about this problem is that they’re actually in a reasonably good situation. Since it can be hard to get a data science job, I would suggest taking any job related to analytics or data. In fact, as I discussed earlier, I got my own start this way. I began to work with analytics and then switched tracks when a data science opportunity presented itself, so I can attest to the viability of this path for entering the field.
Working in the vicinity of data will likely open you to such opportunities. Treat your first job as a stepping stone. Once you get such a job, you should aim to do two things. First, you can lay the groundwork to eventually make an internal shift in the same company to the data science team. By creating good relationships and showing interest in the field, you can establish a good reputation that will help you to pivot. At the same time, continue learning in your spare time, and keep taking interviews.
These were some of the things that helped me both with learning about the field as well as getting a data science job. I hope they work well for you too, and I wish you good luck.
