The unique advantages of non-STEM professionals in tech bootcamps

Suggestions to the new comers to the coding/data science bootcamps

Image credit: Galvanize

Can you introduce yourself?

My name is Ron (Rongpeng) Li. I am a data science instructor at Galvanize Los Angeles campus. Before that, I was a research programmer at Information Sciences Institute, a research branch of The University of Southern California. I studied physics, computer science, and electrical engineering when I was a graduate student.

Why are you interested in teaching STEM subjects?

I have been very enthusiastic about teaching STEM (Science, technology, engineering, and mathematics), especially to people without formal training. At Galvanize, almost all my students come from non-STEM backgrounds. It is well known that there is a gap between the skills, the mindset, and sometimes even the vocabulary used between the tech fields and the non-STEM domain. This may cause students to doubt their confidence and skills. However, I find that the advantages of the non-STEM professionals who transit to the tech domain are less discussed. I would love to bring people’s attention and appreciation to them.

Although I only teach data science, I strongly believe what I am going to discuss applies to other tech domains as well. I will use data science for explaining purposes in the conversation.

What should non-STEM students know going into an educational bootcamp?

Non-STEM students should know that expertise domain knowledge is more important than specific data science skills in the long run. To clarify some possible misunderstandings first, the data skills are definitely important and they are the main reason why students join our bootcamp. However, I find that students have two very different mindsets. Some of them want to change their career completely, leaving their domain. Another group of students want to stay in their domain of expertise but lift their tech skills.

What’s your observation regarding people leaving their career versus people elaborating on their career with new skills?

I didn’t have many students, so I have to observe on LinkedIn as well. My impression is that students who try to completely leave their previous expertise find it VERY hard but the second group may find it more comfortable to tell a consistent story. Honestly, it is not realistic to start in a new domain by simply claiming that you know how random forests work. Sometimes you aim at a scientist job but end up with an analyst job or even no job for a long time. The frustration is pretty detrimental.

What is your suggestion for non-STEM professionals interested in coding bootcamps?

My suggestion is that students shouldn’t think of learning data science skills as a COMPLETE shift of career, at least in the beginning, but a full upgrade of their understanding of their original domain with new skills and perspectives. For example, if you work for a sales team, without data science skills, you are only capable of seeing a fracture of the business. With data science, you can do many other things. This is your story to tell. It is just not possible to get to the music industry to recommend music to people just because you joined the tech bootcamp. For those who really wish to shift their careers, I have two suggestions. First, because your background is always your advantage over other candidates, you should consider starting the transition by finding the intersection of your background and your dream job. It is not hard: for example, if you come from a sales background and are working on software that recommends music to listeners, your software is essentially a sales tactic if you think about it. Why should someone listen to a song? Try to frame your story by using your background expertise.Second, you need really, really good projects to impress people. I would say a WOW-level project. One of such examples that impressed me is a covid-19 infection prediction app. It was very well done.

You mentioned storytelling. Can you elaborate on that?

Sure. This is actually the second point I want to make. I find, in general, non-STEM students have particularly good skills of storytelling. You can say storytelling is a byproduct of working experience, even for people who aren’t natural storytellers.

How do you define storytelling? What is good storytelling?

I define storytelling as a special kind of sales. Galvanize has rigorous graduation requirements, so other than rare cases beyond 3 sigmas, most graduates have roughly the same skill levels. They are essentially qualified for entry-level data analyst or data scientist jobs. The big differences are in the soft skills. For example, some students are very sensitive to numbers and they are visibly excited about increase of model accuracy. However, such students often fail to address the reason why we want high accuracy.

A non-STEM professional with working experience knows the relationship between impact and motivation. You want to make an impact or solve a problem first, and then you bring data science into the room. I had a student who is really into role play. When he gave a presentation, he set up a scene and assigned himself a role. He then created some background information and started from there. This may seem weird, but I sincerely think it is very down-to-earth and necessary. Galvanize data science instructors are also discussing possibilities to bring more business-oriented scenarios into the curriculum like role-playing.

In terms of good storytelling in data science, I think good storytelling requires a sense of the big picture. I think it is also the reason why tools like BI platforms, which are more and more automatic and intelligent,  will not replace data scientists. A good data scientist should know the data aspect well, but should also know the assumptions for the questions being studied, why the data part is necessary, and most importantly, the scope of the possible impact. Including these factors in a data science project makes the project an enjoyable story that doesn’t make people frown. I am not in favor of a particular format of storytelling. Dashboards, reports, animations, or even plain text are all fine options to me.

Sounds like a data scientist needs to be a salesperson. 

Right. To be a data scientist, it is a necessary but not sufficient condition to be a good salesperson.

You have expressed that knowing is not necessarily better than not knowing. What does that mean?

I have chosen my words very precisely. First of all, it is almost always better to know certain knowledge or skills than not knowing them. However, in the bootcamp case, all graduates will reach roughly the same level when graduating. In terms of learning performance, I find sometimes students who are exposed to new ideas for the first time did a better job in terms of understanding and internalizing them.

To provide an example, we did a cross-campus interview to allow instructors from other campuses to interview my students. One of the students said that it was a great idea because the students are likely overfitted to the current instructors. I was shocked. His comments showed that he fully understood the concept of overfitting and can create jokes using the concept. 

I am not saying that years of experience or formal education in STEM is not useful. I am saying that there is no issue (at least, no disadvantages) for non-STEM professionals to step into the domain without prior experience. The screening process at Galvanize DSI program ensures that, if you are admitted, we think you are prepared. If your classmates were in chemical engineering majors, they may have learned linear algebra during their time in college, but they were likely not learning it specifically in the context of something like principal component analysis… but now you may find yourselves in that situation during the bootcamp. Our curriculum teaches everything you need to know in the context of data science.

What are your suggestions for incoming non-STEM professionals?

First of all, bring back the mindset of a student. Learning is probably a long-lost practice. As a student, you will meet challenges. Don’t let the difficulties slip away. Try to manage your time properly so you have time to review and refresh. This will refresh old memories but also build your confidence. Soon, you will find that the dots are connected. Your programming skills influence how well you can understand others’ code and your understanding of machine learning algorithms determines how fast you can translate them into codes. Don’t let this connection make data science hard to crack into but make it a positive feedback loop. The start is always the most difficult.

If you’re looking for more content, I would recommend listening to an episode of the Galvanize Podcast. It is about a full-time mom’s story of transitioning to the tech industry, which is very impressive.

Rongpeng Li (Ron)
Rongpeng Li (Ron)
Business Intelligence Engineer & Author & Speaker

A business insight seeker, an automation geek, a book author and a conference speaker.