How a Ph.D. gets a Data Scientist job
Data Science jobs continue to increase in 2017, yet switching one's job title to Data Scientist can be very challenging like any other mid-career transformations especially from academia.
I think there are 3 parts involved before one lands on a job:
- Gain the basic skill set
- Do projects
- Overcome Catch 22
I will write my thoughts on each part in this post.
Gain the basic skill set
Most Data Scientist positions are posted by industry. The modern industries are increasingly becoming data dependent. More industries are adopting the proven big data infrastructures and are seeking the Data Scientists with the relevant skill sets. You want to make sure you acquire at least a few.
Insight Data Science had a very good blog post to list the skills and how you can learn. Check out.
My case: After I finished a Ph.D. in computational research, I first became a software engineer in data-intensive projects. That is how I became very skillful with those tools.
Good news for people with Ph.D.
Although those skills are crucial in getting things done at an industrial role, the most important skill is missing from the Insight's blog post: Science. Many of people with a Ph.D. training can carefully observe, form hypotheses, and design a study to test. This is a crucial skill to conduct anything beyond descriptive analytics.
You need to demonstrate that you can get the real work done using the skill set. As an industrial Data Scientist, your target audience is business decision makers, not academic researchers. So, choose your project that is relevant to them and present the project according to their viewpoint. Do not write an academic paper or presentation.
Some of the questions you might have to answer as a Data Scientist could be:
- How can we estimate the impact of the price increase of a product?
- How can we predict the cancellation of a subscription service?
- How can we measure the impact of a new product release?
It might be hard to get a data set to run such project, even for answering a hypothetical question. If you are already working in an industry, you may be able to acquire such data by proposing a side project.
The point is that you need to demonstrate that you can deliver actual business values by answering key questions based on your data science skills.
My case: I joined a startup from its early stage where I was excited about the consumer behavioral data we would be ingesting. But we did not have a product that served enough people. So I started as a software engineer, building the products. Four and half years later, we were successful in acquiring users and now we had data with meaningful volume. So, I proposed that our company made a Data Scientists position to help the business make their decisions based on evidence. I did so after running a side project to predict the early cancellation of the client accounts and demonstrated the value of it.
Overcome Catch 22
It is probably easier to get another Data Science position if you already are a Data Scientist in the current company.
There are also many Data Scientist training programs now especially in the US including Master's degree course. Getting a formal academic degree in Data Science or a similar program might be beneficial in getting junior data scientist positions. So, it might make sense for the people early in their careers especially if they are still in the process of acquiring the skills. On the other hand, if you already possess the skills, enrolling in those courses just to get a degree or certificate does not make sense to me considering the time and money you need to spend.
Can you instead first propose a side project in the current position, and present the value of Data Science effort to the company so that you will be the first Data Scientist? Even if you cannot get the official title, your company or manager may assign you to a Data Science project. Even if does not happen, you have a project that you could present to get another job position.
Alternatively, Insight Data Science and other similar program offer an interesting solution for selected talents. Instead of offering course work, they offer an opportunity to work on a data science project, possibly with their partner companies, and the fellows present the project to the partner companies at the end of a relatively short (2~3 months) project. They seem to have high job placement rate in this way.
Despite the increasing demand, switching one's job title to a Data Scientist could be a big challenge. If you are very new to the field, it may make sense to enroll in graduate school or certificate courses. If you already have most skills, you may want to focus on doing projects to show off. It would be much better if the project actually answers business critical questions. This is still a wild west of a job trend and you may want to be bold and creative as to propose your current company to create a job position for you.
Original post: June 7, 2017 | Last updated: June 7, 2017