How to Build a Career in Data Science Coming from a Non-Tech Background

4 challenges and how to address them

Julianna Jia

Published in

Udemy Tech Blog

11 min readAug 18, 2020

Co-authors:

Julianna Jia(https://medium.com/@julianna.jia)

Zhen Xu (https://medium.com/@zhenx)

“I already work in the data industry but coming from a non-traditional tech education background, will this backfire on me?”
“Can I reach the same accomplishment as my fellows who have a technical background?”
“How can I identify my own strength and value?”
“Should I work extra hard to prove myself?”

These questions used to (and still sometimes do) worry us too.

We are both data-industry practitioners with a non-tech background or education. I [Julianna] am a senior data scientist at Udemy, an online learning company. But a couple of years ago, I graduated with a major in education and pedagogy. I developed an interest in data science in graduate school and found that data techniques empower me to better understand how learning happens and build student-focused education applications. I transitioned into a data career in the online learning industry afterward with continued effort.

I [Zhen] am a senior data analyst at The New York Times company. I majored in literature in college and studied educational statistics in graduate school. My interest in data science emerged from my first data internship in the online learning industry when I used data techniques to understand users’ learning patterns. After graduation, I continued to explore data solutions for real-life problems in different industries and I am currently pursuing an online master’s degree in computer science.

We realized that we’re no longer the exception in this industry. With the huge demand for Data Science & Analytics (DSA) jobs [1], there are already many and will be more and more data professionals coming from a diverse educational background just like us. We believe this inclusiveness is particularly important for women who tend to discover their interest in STEM fields later in their life, which allows them another round of opportunities to enter and stay in tech.

We now both have built and advanced our careers in the data domain. However, along the transitional journey, in addition to the questions we shared at the beginning, there are some common difficulties that we met and that might resonate with others:

We lack the same level of training in the theoretical foundation and application as those from institutional backgrounds.
Thus we also experience severe mental stress, ranging from struggling to learn and advance our skills, as well as lack of belongingness to the community.
We also felt confused about the blurry data job titles and the various contexts of data work.
We face the challenge of how to define our own career journey.

Knowing the above and after a bit of research, we found that the tech career for women has a retention problem, especially amongst junior professionals. According to a report by National Center for Women Information Technology [2], 80% of women in the STEM field report “loving their work”, whereas 56% leave their organizations at the mid-level position in their careers.

Additionally, the turnover rate is particularly alarming for women who have 0–2 years of experience in the domain. Only 10% of women leave their non-STEM jobs within 2 years while that number jumps to around 25% for women who work in STEM [2]. Many women leave for the public sector and some of them stay but transition to a non-tech job.

We’ve written this blog post to share our learnings and reflections (below) on how we address these common difficulties faced by junior data professionals with diverse backgrounds. In addition, we also hope to encourage junior women data professionals to stay and shine in their data careers.

Overcoming the imposter syndrome

[Julianna] I didn’t even realize I had imposter syndrome back in my early career until I learned that perfectionism is a typical symptom. I struggled to make things perfect and worked extra hard to earn recognition. But over time, I’ve become much more relaxed and have developed a growth mindset to embrace both success and failure. I give credit to my team on helping me with this. At Udemy, we believe one should bring the whole self to work. My colleagues and managers have created a safe environment for me to be authentic. They assured me it is good to be open and even to be vulnerable because after all, perfection is boring and we are all growing.

Imposter syndrome is traditionally known as the feeling of severe inadequacy and self-doubt. 58% of tech employees report experiencing imposter syndrome [3]. This can be especially true for women [4]. This psychological pattern is likely to happen when you have a new job or learn a new skill and you feel you have to do it perfectly. It is also applicable to early-career women in the data industry who lack a traditional background and are constantly catching up with new skills. What can make it worse is the feeling that no one shares your story or resembles you when looking around. Nevertheless, you may worry that your irrelevant background could hurt your image of qualification or undermine the recognition you deserve.

A common solution to this issue is to hide your true background or act embody the famous Silicon Valley motto, “fake it until you make it.” With this approach though, individuals tend to work extra hard until one day, they either join the established club or more likely, they feel burnout and want to leave.

Instead, we recommend an alternative: embrace your background. You are qualified enough and entitled to your position since you have already successfully gotten your foot in the door bypassing the technical assessments and starting to work on real projects. There’s no need to keep proving to yourself or to others that you’re worth it.

Instead of just trying to adapt, know that you have the skills, strengths, and values to contribute to the team in a meaningful way. For example:

You have tech skills and talents from your training to contribute to the team.
You bring a new perspective from your previous domain, enabling more thinking outside the box.
You connect more dots across tech and society, helping bridge the gap.
You are humble and eager to learn, which fosters a good learning culture.

The idea is, you, along with others, are shaping the data industry together, via your opinions, voices, and actions. So rather than hiding yourself, your openness not only builds your self-confidence but also encourages women who want to enter the tech industry and resonate with your experience.

Keep learning and learn strategically

[Zhen] The learning curve was steep for me when I first started to learn statistics and programming, not to mention learning both at the same time. I took multiple Intro to Python courses until I became comfortable using it without Googling 1000 times every day. The same was true when I was learning statistics. This used to get me so overwhelmed, doubting that I might just not be good at the STEM field at all. This frustration is common. It has taken me a while to understand that learning is an iterative effort and it can be extra hard without having a knowledge map in mind. But it is a solvable problem. Thanks to my background in the education domain, the learning science theory helped me plan out my study, such as connecting the dots for the things that I’ve learned, adjusting the plan based on the project, and most importantly, giving myself more patience to digest and master the knowledge. My reflections are that learning is and should be highly personalized. More importantly, learning ability is developed along with the learning journey.

Lifelong learning is very critical to the data industry, because of the rapid growth of disruptive technology. In addition, learning opportunities contribute highly to workplace happiness[5]. Despite the motivations and rich resources offered by various online and offline learning providers, people from non-traditional backgrounds can feel overwhelmed when picking up advanced skills. Also, it is challenging to keep learning due to the lack of guidance and balance with a full-time job. Besides perseverance and resolution, there are additional ways to think about continuous learnings:

Prioritize your learning with urgency in the flow of work.

Data is such an interdisciplinary domain that newcomers can easily be intimidated by the various areas they need to learn — from coding skills, machine learning, and statistical knowledge to product sense and domain expertise. Hence, it’s important to learn the most urgent, in terms of the skills needed immediately for the working task, to help prioritize your daily learning[5]. The bonus of such prioritization is that you are learning by doing, which is proven to be better than only watching videos and reading books[6].
People are also more motivated when doing task-driven learning because the result is reflected directly[7]. Applying your learning on a real working task helps you quickly internalize and test out the theories you learned, thus forming your own opinions and experience of a domain.

Learn iteratively since learning is not a one-time battle.

Similarly to the best engineering practices, learning is also an effort of iterations as “the deepest ‘aha’s’ spring from an encounter and then a return” [8]. Learning gradually and sequentially is important. To learn a new technique, you may start with introductory videos, followed by practicing with Jupyter notebook tutorials. After adequate practices in the domain, you start to read relevant academic papers and books about a niche area as needed. Your interests and needs to decide where you start and pause in the journey. Learning from breadth to depth also matters.
To internalize the knowledge, you may have to revisit the same concept over and over within 1 or 2 years across projects. However, learning iteratively is not a simple repetition. Instead, each iteration builds on the previous one, with new use cases and different data, more in-depth research and trials, and discussion with co-workers. The understanding of such concepts becomes granular and also the knowledge map is formed with more dots connected. After all, forgetting and relearning is common. There’s no perfect learning path. We are all Sisyphus on the journey of learning.

Courage and self-confidence outweigh intelligence.

Learning is not only an intellectual activity but also a constant emotional rustling. Do you constantly worry about your ability to learn before you even start? Stop doing that, since mental burden only adds frustration to the learning and memory system[8]. The aim of learning is not to prove you are smart, but to acquire the skills or knowledge you desire. So the best strategy is actually just to learn it.

Create belongingness

Belongingness is not only a feeling you can seek around but also an environment you can create for yourself. For example, we met in an online mentorship program two years ago (Julianna mentored Zhen for 6 months) and became good friends after the program ended. Instead of a one-way offering, we found the mentorship is mutually beneficial from the very beginning. Since we share similar working context and personal interests, we not only discuss technical problems and career development, but also share personal struggles, vulnerable feelings, and support each other in many ways. Though we haven’t met in person, this feeling of belongingness, that someone’s there to listen to you and share your feelings, encourages both of us to travel further in our careers. We also learn that belongingness is all about sharing and empathy.

According to a study, the workplace experience, which includes belongingness, matters for women to stay in computing[2]. Hence, it is important to form such belongingness for young data professionals from non-traditional backgrounds. There are multiple ways to create a supportive community:

Become a mentor or a mentee.
Form or join inclusive reading and learning clubs.
Share and learn from experiences at daily work.
Offer help and ask for help.
Give and receive praise and constructive feedback
Speak up (by writing blog posts, making podcasts or via other mediums) and listen

You can find followers and pioneers in these communities, who make you realize that you’re not alone in the darkness. Last but not least, we should seek both internal and external recognizations via better self-awareness, self-promotion, and advocating on behalf of the community, to fight institutional barriers and establish belongingness together.

Define your own path

As an innovative career field, the career path in data is full of opportunity as well as ambiguity. The blurry landscape of data jobs often doesn’t match the job title. For example, “data scientist” can stand for an analytics role in some places or machine learning engineers in others. Not to mention the career growth choices range from technical positions to business positions, including business intelligence engineer/analyst, data engineer, machine learning appliers, research scientist, technical manager, product manager, etc. Given the interdisciplinary domain, we can define our own path that follows our interests and strengths.

It’s important to explore your real interests and strengths in your daily work. Ask yourself what fulfills you more — researching, engineering, or informing strategy. Exploit your work with your interests and pave your career towards your inclination. Instead of comparing yourself to others and trying to improve your weakness, identify and amplify your strengths. Your career is as broad as you explore.

Both of us changed multiple roles to advance and grow our careers. I [Julianna] joined Udemy as a growth team data analyst 5 years ago and discovered my interest in building data products via algorithms later. The growth mindset culture of Udemy encourages me to explore different data roles. I first did rotations at the data engineer team and data science team to test out the experience and then switched to the data science track to work on recommendation data products. What I learned from this experience is that a career path is not a paved road, but a journey followed by one’s passion.

Similarly, [Zhen] I benefited from the flexibility of working at the NYT and had the opportunity to dive into all aspects of business problems. I started with the subscriber analytics team, which focused on a/b testing and dashboarding and then transferred to the advertising team that builds data products. I believe that the key to owning your career is to navigate different roles, find your strength and the field that interests you the most.

Conclusion

Getting into the data industry is not the end of the journey. Remember, you’re not alone at any stage of your career. Keep learning, believe in yourself, help each other, navigate, and shine.