First of all, I want you to know that I am the perfect person to explain how to switch your professional career towards data science. The reason for me to be so sure about that is because I myself went from zero to hero as a data scientist after successfully shifting my career.
Back in 2016 I was working at BMW as a mechanical engineer in the development department. All the tasks I was doing back then were purely mechanical. I hardly ever dealt with data (beyond data visualization). In 2018, I wrote my MBA master thesis in the field of Smart Grid. That thesis opened my eyes. The amount of challenges and therefore opportunities in the field of data science were endless (I’ll talk about that in future posts). I did not hesitate, after handing in my master thesis in August 2019, I quit my job and fully shifted my career towards data science.
In this post I am explaining, in a structured way and at a high level, how to successfully and efficiently switch your career towards data science. I believe there are three steps to achieve that: get informed, assess your skills, and find your path.
Step 1: Read, read, read
I guess it was the influence of the MBA that, back in 2018, the capitalist side of me was emerging. Even though I had a decent salary, I wanted to make more money, so I remember spending a few hours every now and then googling “how to make money on the side”.
It was then when I discovered the concept of “data science”. I could not believe what I was reading: “sexiest job of the 21st century”, “data is growing exponentially, and with it the demand for data scientists”, “4th industrial revolution”, “up to 150.000$/year”. Was that for real?
Abandoning a promising career as a mechanical engineer after 8 years was not something you decide in a couple of days. To be sure about such a decision, I had to learn more, read more. I was searching things like: “Is it too late to shift your career at the age of 30?”, “Does data science have a future?”, “Is it better data science or computer science?”, “How to shift from mechanical engineer to data science?”, “What skills are needed in data science?”… Having a big picture is of extreme importance to make a proper decision. Each context is unique, and so are the actions one must take in it.
So analyze your context, ask questions about who you are, where you are, what are your resources, what do you want, why do you want it. You need the big picture for the right decisions on how to shift your career towards data science.
Step 2: Assess your skills
In a way, in this section we are still building the big picture aforementioned but now deep diving a bit into technicalities.
You have probably heard about the mathematical, statistical and programming skills needed for data science. You have probably got scared too by them. Well don’t be.
I am not sure where you want to get with data science. Do you want to work for Google or Amazon Alexa and improve their Natural Language Understanding algorithms? Or do you want to work in a small-middle sized company first and then see where it goes?
See the following analogy (that’s the way I think, it’s my logic, it might not be accurate): 90% of companies worldwide are SMEs. These are companies with 250 or less employees. The small companies are mainly going to offer services to the middle companies, and most of the middle companies are going to offer services among them. SMEs do not have the budgets of massive companies such as Google and Amazon. Therefore, they are going to solve problems using “mainstream” data science. This means using methodologies and computational power reachable by all companies. If I say this, it is because for 90% of the cases, it will be enough for you to have the skills of a “mainstream” data scientist. You will not need to go many extra miles as you would to work at Amazon or Google.
That being said, the following overview is for those who want to start as a “mainstream” data scientist, and therefore be capable of working in 90% of the companies worldwide. Following, the minimum skillset I believe is needed to land a data science job:
- Programming: 6 months of Python programming. Yes, there are other programming languages, but with this one you play safe and easy.
- Mathematics: What you knew from high school. You need even less than that. I just one you to relax because the scope of mathematics needed is not much.
- Statistics: The exact same thing as mathematics.
- Machine Learning: You need to understand how machines learn and some of the algorithms used for it.
- Industry domain: This is not a skill but it will make the whole transition simpler.
I am not going into detail in this post about the maths, statistics and machine learning needed. Just keep in mind the high school thing I mentioned. In regards to the domain, I believe you should, at least at first, choose an industry or topic you want to build your data science skills upon (for instance in my case the domain is industry 4.0 and energy because of my mechanical engineering).
Now it is time to assess your skills. I suggest you to draw on a paper a radar chart with the 5 skills previously mentioned and, on a scale of 1 to 10, mark where you are today. In the following image you’ll see my chart before and after shift:
This was my train of thought to fill the chart:
- Programming: I programmed in the university 15 years before the shift but I have not touched a piece of code ever since, therefore 0.
- Mathematics: I was good in high school (I’m also a mechanical engineer). Therefore 7.
- Statistics: I used to be good at the university but I forgot almost everything, therefore 3.
- Machine Learning: Never heard of it before, therefore 0.
- Industry domain: I had a really strong domain in many industries and functions, therefore 8.
Now that you have this technical view of your skills, you will have a clearer idea on what path to choose.
Step 3: Find your path
Now that you have a clear idea of your context and what your skill set is, it is time to find the best path for you to shift towards data science.
I identified 4 different paths you can follow depending on your context and skillset:
- Bootcamp: Intensive, three-to-six-month programs that prepare individuals for entry-level and junior data science jobs.
- Degree/Master: You can upgrade your skills and become a data scientist through the university (that’s the path I chose).
- Self-taught: All the information you need to become a data scientist is online, of course there are
- Shift within your company: Some companies will facilitate your transition to data science within the company.
Note: I did not include the option of becoming a data scientist by means of a PhD because this entails many other factors and complexities.
Now, in order to choose the right path, and keeping in mind your skill set and context, I believe the following parameters need to be considered:
- Length: How long will the path take you to complete? Do you have the time and endurance to go for it?
- Cost: Of course the paths are radically different in terms of price.
- Mentorship: Really important. It is not the same to have professors guiding you than solving the problems yourself.
- Flexibility: For some paths you need more compromise than others. This refers to how intensive the length is.
The following table shows a high level classification of all paths considering the dimensions. Prices are again part of your context, I can not specify them. You must do your own research:
Both universities and bootcamps will make sure to assess your skill set and maybe give you some preparation homework prior to the course. As a self-taught or within the company this will be more tricky, you’ll need to find some sort of mentor to guide you. Preparing a project portfolio is also of utmost importance, but I’ll talk about that in future posts.
In my case, I chose a one year long part time master in data science. My context was simple: I could not do transition within the company, I did not want to interrupt my career to do a bootcamp, self-taught was going to be too slow considering my goals, master was the remaining option and I was happy with it: PhD professors teaching the latest data science trends, structured classes with homework and corrections, creation of a final project for my project portfolio, and I had the money for it!
One last comment before the end. Regardless of the path you choose, I recommend you to first do one or two months of online courses to get a sense of what data science is. For instance, I did a Data Science path with Python in datacamp.
Please don’t forget to register to the newsletter (see on the right) for further tips in data science, and to follow my founding journey on Instagram and LinkedIn (see on the left).