Official Blog
learning data science
+2

Applying for data science jobs and how to set yourself apart

Learn how to apply for the jobs you desire and set your self apart. Learn what makes you stand out and what steps to take stand out from crowd.

Introduction

In 2012, Harvard Business Review named data scientist as the sexiest job of the 21st century. Ten years later, despite the emergence of AutoML platforms, (e.g., AWS Redshift ML and Google Cloud AutoML) and the pandemic-era slowdowns, data scientist remains one of the most wanted roles around. In fact, the data scientist ranks the highest paid role in the tech industry with an average yearly salary of $150,000.

Although their specific responsibilities may vary from industry to industry and from company to company, most data scientists share the mission to help organizations create value from data. They do this by exploring patterns and trends in a large amount of data, communicating results to a broad range of stakeholders, and building and maintaining models to enable automated decision-making. Hence, becoming a data scientist requires a unique, diversified skill set covering statistics, coding, business sense, and communication—all to be demonstrated through a series of interview questions and take-home challenges.

With this blog post, we would like to share a few tips, from the perspectives of both an applicant and an interviewer, to help you stand out in a data scientist interview and get your dream offer.

Tip #1: Get Familiar with the Role and the Company

As mentioned above, “data scientist” is a vague term that can refer to any roles that revolve around data. Two data scientists from different companies or different industries may find themselves engaged in completely different types of tasks. Hence, it is essential to read through the responsibilities section in the job description, or ask the interviewer, “What would a day in this role look like? Would this role spend most of the time exploring and visualizing data, or building models?” The more specifics you learn about this role, the sooner you will learn whether this role matches your profile and interests. This helps you target the positions that are a better fit and increases your chance of landing the job you truly want.

You could find useful tips on how to improve personal branding and profile matching on career websites like indeed and linkedin. If you are looking for career advice tailored for data science and analytics, the career services of Datacamp would be very helpful - here you can find personalized sessions from career coaches who are specialized in this area.

Once you have targeted the right role for you, the next step is to show the company your passion for their business. This can be a little tricky for those who are just starting off as a data scientist or those who want to switch industries, as you may not have much experience in solving business questions in the industry that you’re applying for.

One thing we find really helpful is to work on projects that are relevant to the target industries and mention them in the resume and interview processes. For example, if you’d never received training in finance but wanted to apply for a data scientist role at an investment bank, working on a guided project like this one Modeling the Volatility of US Bond Yields on DataCamp can help you learn what kinds of business questions a data scientist working in finance is expected to solve.

Kaggle is also a good platform where you can find a lot of interesting competition and datasets. By working on these projects and talking about them during the interviews, you will not only build basic knowledge of that industry but also show the company that you are so interested in their business that you are spending your spare time exploring their use cases.

Tip #2: Let Your Technical Expertise Shine Through the Take-home Challenges

Usually, the hiring process of data scientists involves a take-home challenge where candidates are given one or multiple datasets, and a few business questions to solve. Although the submission requirements may vary, the candidates are usually expected to share the code, models, and analysis output.

Many candidates see this as a great opportunity to showcase their technical skills. R and Python are the main data science languages and both have their own must-have packages for data wrangling, statistical modeling, and machine learning e.g., pandas, sklearn and statsmodels for python, as well as tidyr, dplyr and caret for R. If you are new to data science, it is recommended to take a few career track courses like these on DataCamp: Data Scientist with Python and Data Scientist with R. In addition to an extensive range of courses, Datacamp also provides a workspace where one can practice with pre-written code templates and pre-configured datasets for those who are looking to gain some hands-on experience to close the learning-doing gap.

What can sometimes be underestimated is the role of data storytelling. Good data scientists are also effective storytellers who can communicate their model outputs well enough to convince stakeholders and thus drive real changes. As the saying goes, “a picture is worth a thousand words”. Therefore, make sure you include intuitive data visualization in your analysis report to help interviewers understand how you uncover the patterns in the data, as well as show how your model can create value. In addition to common packages such as matplotlib (Python), seaborn (Python) and ggplot2 (R), plotly is a graphing library worth trying if you want to build a web-based, interactive dashboard. This course on DataCamp could be helpful: Introduction to Data Visualization with Plotly in Python.

Alternatively, you may try no-code BI tools such as Tableau, Power BI, and Google Data Studio. These tools give you two advantages. On one hand, they are popular visualization tools that are used by most non-tech stakeholders such as data analysts and product managers. Hence, mastering these tools shows that you can easily integrate with the data analytics stacks that the company uses. On the other hand, they provide more customization than Python or R packages do, which allow you to easily build a slide-style, interactive analysis report. If you are new to these tools, this Introduction to Tableau course on DataCamp is a good starting point.

Last but not least, the delivery of your code also matters. Unless otherwise specified, it is always a good idea to build your solution on Git and have all dependencies readily incorporated into your code. Nothing creates more frustration for interviewers than when they receive a zip file and try to run the code, yet only see error messages like “package XYZ can not be found”. Meanwhile, make sure your code comes with clean, proper documentation which makes it easy for people to follow your train of thought.

Tip #3: Keep in Mind the Data Product Lifecycle when Sharing Your Previous Project Experience

Most of the time, candidates applying for a data science job are expected to share their previous project experience. A common pitfall is to spend too much time describing the technical effort e.g., how they clean data and fine-tune (hyper)parameters. This might lead to overloading your interviewers with the information they may not fully understand, and at the same detracts from the limited time they have to discover your business acumen and stakeholder management skills.

Hence, I would recommend that when preparing for the interview question about your previous projects, make sure your answer follows the STAR framework and covers all stagesof the data product lifecycle, even if you are not directly involved in all of these areas:

  • Business goals and questions: What business problem(s) does your project try to solve? What goal(s) or KPI(s) does your project contribute to?
  • Data collection: What challenge(s) did you have in gathering the data and how did you tackle them?
  • Exploratory data analysis: How did you present your analysis result to non-tech stakeholders? What questions or criticisms did they raise and how did you respond to them?
  • Modeling and logic: Why did you choose the model(s) you finally implemented (discuss both technical and non-technical motivation)? What are the key limitations of your approach?
  • Testing and deployment: How did you test and deploy your model(s)?
  • Monitoring: What metrics did you choose to evaluate the model performance? What learnings did you have and how did you improve the model based on your learnings?
Figure 1. Data Product Lifecycle

Moreover, don’t forget to highlight the different roles that were involved (e.g., product managers, data analysts, data engineers, QAs, and business operation managers), their responsibilities, and how you interacted with them. Hearing this, your interviewers are likely to be impressed with both your business sense and teamwork.

Conclusions

With this blog post, we looked at the recent growth in the demand for data scientists. We then discussed a few tips which can help you crack data scientist interview questions—both technical and non-technical. If you aspire to land a data scientist job in the future, you may be interested in trying out these tips or building your way toward your dream offer by starting with an accredited data scientist course.