Data Science Is a High-Paying, Fast-Growing Field—Here’s What You Need to Know About Entering It was originally published on The Muse, a great place to research companies and careers. Click here to search for great jobs and companies near you.
Ever wonder how Instagram knows exactly what ad to show you? Or how your supermarket keeps the right food in stock? It’s probably thanks to a data scientist.
You may have heard the words “data science” tossed around a lot in the last few years. It’s a high-paying and swiftly growing field—with demand for data science expertise in every industry and pretty much every type of organization.
But what exactly is data science and what does working in this kind of role entail? As a data scientist myself and the founder of for the love of grad—where I help clients get accepted and funded at top PhD and MS programs—I can walk you through what data scientists do, what it’s like to work as a data scientist, how you can become one, and why you should consider it.
Data scientists write code to model and answer business questions. Take those Instagram ads, for instance. Using all of the different data points collected, a data scientist would create a model or algorithm that’s general enough to be applied to each user but specific enough to produce relevant recommendations for which ads to show to each person and when. This is no small feat.
The amount of data that is generated in the world each day is enormous, and it’s growing exponentially. We can’t use just Excel or similar tools to wrangle all this data; instead a data scientist uses programming languages, the most common being Python, to make sense of the data.
You might be thinking, OK, so you sit down with all the data and throw it in a magical machine to spit out the best answers to the question you’re trying to answer, such as, “What ad should this user see right now?” Not quite. A data scientist has to create that magical machine or model—one that’s tailored to what their organization is trying to achieve.
The process starts with asking the right questions. “Perhaps the biggest skill of a good data scientist is critical thinking,” says Kate Lyons, a senior data scientist at PPG Industries. Indeed, simply throwing data into some premade algorithm won’t help you make a meaningful impact at work. It’s important to break down the problem and ask the right questions, or you’ll end up solving a problem no one cares about.
Once you have a specific question, you then conduct data exploration to identify initial patterns and trends, and finally you begin experimenting with different models. “You have to really understand the methods you’re using so you don’t make an expensive mistake or inflate results,” Lyons says. Data scientists know which types of models generally work for what kinds of problems, but it’s still an art to make the best model possible—for example, a model that says, “At 8:05 PM, show Vincent an ad for sunglasses” (thousands or millions of times over for each user at various times).
When you’ve come up with what you think is the ideal model, you need to work with the project team and stakeholders to make sure it provides the value they’re looking for. This brings up another crucial aspect of data scientists’ roles: deploying the art of storytelling. “You should be curious about the data and the stories you can tell with the data,” says Abhilash Nair, a lead data scientist at FICO. You have to be able to frame your stories in a way your audience (senior leadership, colleagues across the organization, and anyone else) can understand. Armed with your shiny new model, your story, and your ability to finish a task imperfectly (a model will never be perfect right off the bat), you then convince stakeholders this is the best way to go given the constraints.
Finally, the product development team will package the code you’ve written into a product that either solves an internal business problem (What will the demand for this brand of cereal be next week?) or improves a customer-facing application (What ad should be placed where in a user’s feed?).
Though exact tasks and responsibilities can vary, most data scientists:
- Help frame business questions so they can be answered with the available data
- Do data detective work to figure out what data is useful and/or corrupted
- Work with data engineers to ensure the right data is available
- Write code to model data
- Work with project teams to sell the model and its business value
- Learn new frameworks and tools almost daily
Data science is a broad term, and there are many job titles that fall under the bigger umbrella. These are some of the most common (click on each title to search for open roles on The Muse!):
- Data scientist: Creates and codes machine learning or artificial intelligence models to answer questions and make predictions using big data
- Machine learning engineer: Does the same work as a data scientist
- Artificial intelligence engineer: Does the same work as a data scientist
- Big data analyst: Does the same work as a data scientist, but may work on more visualizations rather than AI
- Pricing strategy data scientist: Works on a subset of data science that focuses on setting prices for products and services
- Risk analyst: Uses statistics and coding to assess risk with big data sets
- Data engineer: Builds pipelines to extract, ingest, and warehouse data
- Cloud data engineer: Creates the infrastructure used to run different data pipelines in the cloud and may or may not do data science
A data scientist can wear many hats—and the lines often get blurred. But the heart of a data scientist’s day-to-day work is writing code and creating models. A data scientist may also work on front-end visualizations and user interfaces, but usually these tasks are better suited to software developers.
What you do is important, but so are where and how you do it, along with what salary and job prospects you can expect. I may be biased, but being a data scientist offers great options that appeal to almost any analytical thinker.
Where Do Data Scientists Work and What’s the Environment Like?
Data scientists mostly work in office environments across all sectors (banking, automotive, tech, healthcare, retail, etc.) at any type of organization (from small startups to giant corporations, international nonprofits, and more), which means there’s a huge range of work environments and cultures. The big tech hubs—Silicon Valley, New York, Seattle, and Boston—harbor many of the country’s data scientists, but in the war for talent, data scientists are often given the flexibility to work remotely.
The type of organization you’re hired into could drastically alter your role as a data scientist. For example, some companies have central data science teams that act as consultants, helping different business groups make sense of their data and creating compelling models to support business strategies. In that type of role, you might have to speak with various stakeholders, assess use cases, explain the best approach, and then go back to your desk to build the model before handing it off for deployment. In another scenario, at a large tech company, you’re more likely to be on a very specialized team, focused on optimizing specific parts of models to do one particular task. When you interview with different organizations, be sure to ask about the typical tasks for that role, as they vary widely.
Can Data Scientists Work Remotely?
Heck yes. Data scientists need a computer and access to data. That’s about it. So it’s technically feasible to work from anywhere if you have those things. In practice, it all depends on the company. Many companies are advertising remote roles in an effort to woo data scientists. Others prefer in-person brainstorming sessions. So whether you prefer remote or in-person, you definitely have your choice.
What Hours Do Data Scientists Keep?
Data scientists usually work standard business hours, between 8 AM and 5 PM, Monday through Friday. Some organizations are allowing more flexibility where core working hours (say, 10 AM to 3 PM) are observed, and beyond that employees are given the freedom to work different schedules. A standard 40-50 hour workweek is most common for data scientists, as is a good amount of autonomy. “I typically work 9 AM to 5 PM, but I’m fortunate and privileged to work for a group that understands flexibility,” Nair says. “No one is micromanaging how or when I work.”
What Are the Job Prospects Like for Data Scientists?
Data scientist has been touted as one of the hottest jobs out there since 2012. Nearly a decade later, you might be wondering whether the market has been or will get saturated. The answer is no. According to the U.S. Bureau of Labor Statistics (BLS), data science is one of the top 20 fastest growing fields, expected to grow 31% between 2019 and 2029. In comparison, the average growth rate for all occupations is about 4%, meaning data science is projected to grow at almost eight times the average rate.
How Much Do Data Scientists Make?
Not only is the data science field growing, it’s also one of the highest-paying jobs on the market. BLS data shows that the 2020 average salary for data scientists was $103,930. And the earning potential goes up as you gain experience. While the salary for data scientists ranged from $67,000 to $135,000 as of May 2021, according to the compensation resource PayScale, salaries for senior data scientists ranged from $95,000 to $161,000, those for data science managers ranged from $96,000 to $180,000, and those for data science directors ranged from $117,000 to $203,000. One report from the recruitment agency Burtch Works found that high-level data science managers made an average of about $250,000 a year.
The two biggest skills a successful data scientist needs are:
- The ability to understand, break down, and solve complex problems with code
- The ability to use data to tell a compelling story
If you’re an excellent coder and are waiting for someone to tell you exactly what to build, this probably isn’t the job for you. If you want to talk strategies without ever writing a line of code, you’re also in the wrong place.
Data scientists are indeed scientists. They must structure a problem into an answerable question, form a hypothesis, set up an experiment (code and model), and assess how well the model answers their initial question. They don’t know if it will work. In fact, it often doesn’t. The ability to dig deeper, try new approaches, and persevere make a good data scientist great. At the same time, it’s important to remember that you are “working in a business context, and your job is not to endlessly dive deeper, but rather solve a business problem to provide value,” Nair says.
Depending on your role, you may also need excellent communication skills. Initially, you may need to work with multiple stakeholders to define the problem. This is not trivial, as many business users are not knowledgeable in data science methods. You must be able to translate abstract business desires into technical requirements that can be accomplished with code. Then, once the model and code are complete, you will likely present your results to decision makers. Your ability to sell the merit of your work is paramount in moving forward with your project and ultimately getting your model into production. As Lyons’ manager once told her: “You can be the best data scientist in the world, but that doesn’t matter if no one understands what you did and why you did it.”
Almost all data scientists have a BS, and many have master’s degrees or PhDs. In fact, according to 365Data Science, 80% of data scientists have an advanced degree. And many data science positions require at least a master’s degree.
However, you don’t need a specific “data science” degree. Data scientists can come from diverse backgrounds, such as a PhD in neuroscience or a BS in mathematics.
Generally, there are a few routes to becoming a data scientist:
- Complete a B.S. degree in computer science, computer engineering, statistics, or mathematics. Learn typical data science frameworks (Python, PySpark, AWS, Azure, etc.) as part of your undergraduate coursework or through side projects. Get a job as a data scientist.
- Complete a BS degree in any field. Complete a data science bootcamp (there are many). You may be able to get a job as a data scientist right away, or you may start as a data analyst and work your way up.
- Complete an MS degree in data science, computer science, or related field. Learn typical data science frameworks within your program. Get a job as a data scientist.
- Complete your PhD degree in almost any subject area that involves analyzing data (bioinformatics, political science, astrophysics, etc.). While completing your PhD, learn typical data science frameworks—you’ll need them to analyze your data—and try for internships to learn “real world skills.” Upon completing your program, get a job as a data scientist.
Do you see a pattern here? You need to somehow learn to write code and solve problems using data. However, your academic background may affect how difficult it is to break into data science. In my experience, more weight is given to those with computer science or math degrees, and those outside STEM fields generally need to have more experience to offset the alternative degree path.
Everyone wants to claim they’re a data scientist, but you have to actually back it up with skills and experiences. That’s where specific degrees or bootcamps come into play. Due to demand, universities and all sorts of companies have begun offering various ways to get data science credentials, including online MS programs, nanodegrees, bootcamps, and certifications (such as AWS or IBM). When figuring out which road to take, make sure you ask about job placement and networking opportunities.
Additionally, no matter what your background, having a portfolio on GitHub could also help showcase how you’ve used your skills in practice to do real projects.
If you like solving problems, love to code, and are at your best when you’re always learning something new, this job is for you. If you’re afraid of coding and want to be working with people at all times, this isn’t your jam. Data science is completely woven into all major businesses, and the ability to analyze data is a prized skill. So if you’re looking to work across organizations all the way to the C-suite and have an impact, then consider pursuing a career in data science.
There are a few other things to note as you decide whether data science is the right path. Diversity in tech is not great. Data science is no different: Roughly 20% of data scientists are women, 5.8% are Black, and 3.3% are Hispanic or Latinx. Which means if you’re a woman or person of color, you might find yourself “the only” on your team. That can be a difficult experience, so it’s something to consider going in.
If you toss the less-than-stellar diversity statistics in with the ability to bend data and code to your whim, it’s also no surprise we’ve seen unconscious bias seep into the artificial intelligence (AI) models that data scientists develop. But if the lack of diversity and biased algorithms alarms you, I urge you to get off the sidelines and become part of the solution—if you’re helping create the models and write the algorithms, you can help advocate for more equitable ones.
If you’ve had your eye on artificial intelligence and data science, you might know that there are many automated AI solutions available these days, such as H2O.ai, einblick.ai, Azure autoML, and AWS Sagemaker. So if machines are just going to take over, why should you pursue a career in data science?
Well, the machines haven’t taken over yet. Frankly, software development has been in the automation phase for years and yet more and more specialized skills are needed to keep up with the mounting complexity. Likewise, we’ll continue to need data scientists and the skills they bring to solve complex business problems.
In fact, due to this high demand for data scientists, we’re seeing a new role emerge: citizen data scientist. A citizen data scientist is a person who doesn’t have extensive training in data science, but does have deep domain knowledge and can turn to easier-to-use tools to create data analyses and models. “Companies are drowning in data. We need to enable people with domain expertise to do more on their own, and solve their own problems,” says Tim Kraska, a professor at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) and CEO of Einblick Analytics. “For example, DARPA [the Defense Advanced Research Projects Agency] is looking to have a data scientist in every operational unit, but there aren’t enough data scientists to go around.”
This is good news for people looking to upskill within their current organization. But what does it mean for the trained data scientist? Will they become obsolete? The short answer is, once again, no. While a citizen data scientist may create first draft analyses and models, a trained data scientist may be needed to make them more accurate and deployable or serve as an expert in a particular class of problems. Data scientists may also be instrumental in training and upskilling colleagues across their organizations. In short, data scientists will continue to have a prominent role for years to come.
Whether or not you decide to dive deep into data science and make it your career, data skills are a necessity in the modern workforce. So even if you’ve read all this and realized it’s not for you, you can’t go wrong learning more about how to use data and solve problems.