The world’s leading publication for data science, AI, and ML professionals.

Data Science Portfolio from Employer Point of View

What is the company looking for from your Data Science Portfolio?

Photo by Kelly Sikkema on Unsplash
Photo by Kelly Sikkema on Unsplash

If you enjoy my content and want to get more in-depth knowledge regarding data or just daily life as a Data Scientist, please consider subscribing to my newsletter here.

In recent years, companies realize how important it is to utilize their data to get an edge over their competitor. From corporate, startups to small businesses already start to use data projects to improve their business. This, in turn, increases the need for qualified data people – something that until now still keeps increasing.

With the hot industry is rising and the need is increasing, many people trying to learn Data Science to get into this field – which creates fierce competition. Why thou that the needs increase, but the competition is still fierce? Because the field is still new, and many companies are looking for experienced individuals rather than fresher ones. That is why the data science position entry position becomes the bottleneck. So, how do you stand out in the sea of applicants? One way is to tweak your portfolio to meet what the employer wants.

In this article, I want to outline my experience of what I am looking for from the potential applicant Data Science Portfolio. Let’s get into it.


Structured Portfolio

What is more unpleasant than seeing a messy data science portfolio? The story is not clear, and the conclusion is not present – that is what happens if you leave all your coding and data exploration portfolio as it is. Messy project is a problem that I often see: People already know what kind of data science project you want to do and try to solve; however, they did not create a structure from your Portfolio that allows people to digest the process.

For example, an individual wants to solve the fraud problem by creating a model to predict fraud and achieve good metrics. The individual managed to secure an interview because of this fraud model but did not form a good structure story of how the model developed and would help the company. This individual did not get an offering because the interviewer feels the individual cannot clearly explain the workflow and was afraid that this individual would be a mess during the employment.

This example is something that commonly happens in my experience. People come only to focus on their results and not elaborate the process – no structure at all. The employer wants to see not how good the model you create but how your thought process is to achieve this model. It is important to create a structured data science portfolio to stand out from the others.

The structure could be varied, depends on your own Data Science project. However, it should always tell a story from the beginning to the end – even if it is in your Notebook. The common structure is:

  1. Explain Problem you want to Solve
  2. Explain the data and how you acquire it
  3. The data exploration
  4. The model process
  5. Result and Conclusion

With these five steps, you could already create a nice structured of your data science portfolio. You could create it using presentation, Git, Notebook, or any Mediums – as long as it is telling a good story. Extra point in my opinion if you could explain the detailed benefit by applying your project.


Address the Business Problem

It is a great feeling to develop an amazing neural model that could accurately predict the facial features and clothes that someone is wearing, but would this project solve any real problem that the company are you applying to have? It might be, but with many stretches – which is unlikely to be at the current interview stage.

The company hire data scientists to solve their business problems where each company would have different problems. The job ad would often only list the technical requirements (with some domain-specific) but rarely spell out what kind of problem they need to solve. When you are applying for this kind of job with a generic portfolio, it might work, but it might do not; because many applicants would have submitted a "similar" resume and data science portfolio.

How to stand out then? Be what the company want! And that is to solve their business problem. Try to create a specific data science portfolio that addresses the business problem; for example, you apply for a data scientist position in the financial industry – this means you could create a data science portfolio that addresses the common problem in this industry, such as investment, fraud, risk, and many more.

I love to see the applicants who, at the very least, research about the company they are applying to because the research they do means that they are willing to spend extra time knowing the company. It is certainly in the company best interest to employ someone who already understands the business or the company compared to the one who did not have the knowledge – which is why it would make you stand out if your data science portfolio addresses the company business problem.


Being Creative

This is related to how your data science portfolio should address the business problems. When you decide on your project, do not use a staple project that you learn from the courses or book – rather, you should build the data science portfolio based on your own creativity.

Imagine that you are applying for a data scientist position in the bank industry with the titanic or iris project; what it would look like?

First, it is a project that everyone did.

Second, no business in need for those projects (maybe historians or biologists?).

Third, it would undermine your chance because the employer would think you are lazy.

You could do many other projects to show how creative you are when building the data science portfolio. The creativity could be shown in many ways, including:

  • What business problem do you try to solve
  • How you approach the problem
  • The way you are exploring the data and summarize the insights
  • Developing the models and troubleshoot the technical issues
  • Creating a conclusion of your data science portfolio

and many more. Creativity means you are not limited to the courses or the common practice, but how you can break through from it.

From my experience, when I am exploring the data science portfolio from the applicant, I would try to look at how they solve the problem they decided to solve. Many people tend to fall in a similar pattern, but when I find someone who could give me different things (In a good way), I will become excited.

I remember many applicants who use a similar dataset as their data science portfolio and try to solve the same problems; however, only one stands out. He is stand out because the way he summarizes the insight is different. Many people summary the insight as the features is not important. Still, this one individual manage to do a deeper analysis and find that the actual problem lies in the compounding problem – which needs creative thinking to know that this is the problem.

In a way, being creative might need more practice because it is related to critical thinking. Still, once you get used to seeing the problem from many sides, I am assured that many employers would crave your creativity.


Conclusion

Data science is an industry that is still rising and with high competition. To successfully achieve the position, you would need to have a good data science portfolio. However, what is the employer looking for from your portfolio?

In this article, I explain my experience from the employer side as what I am looking for from the data science portfolio. They are:

  1. Structured Portfolio
  2. Address the Business Problems
  3. Being Creative

I hope it helps!


If you are not subscribed as a Medium Member, please consider subscribing through my referral.

Visit me on my LinkedIn or Twitter.


Related Articles