Abe Gong says coding is a part of his identity — even though it’s not a part of his current role.
“A lot of data people start off as a coder,” Gong, founder and CEO of data quality and collaboration company Superconductive, said. “You spend a lot of time hands-on-keyboard. Along the way, you get very good at understanding the business, storytelling and working with other stakeholders. The skill set ends up looking a lot like management.”
And therein lies the tension, he said. Career advancement can often mean moving into more managerial roles and can translate into less time doing the hands-on work of coding and interacting directly with data — what many data scientists love and what drew them to the field in the first place.
But that doesn’t have to be the case. Here are some strategies data scientists can use to stay hands on as they advance in their careers.
5 Strategies to Keep Coding While Advancing Your Data Science Career
- Pursue a technical track. See if your company offers a technical track in addition to a managerial track for career advancement. This would allow you to continue hands-on coding as an individual contributor.
- Specialize in a domain. By being a technical specialist, you are less likely to be drawn away from coding and directly working with data.
- Stay at a company to develop institutional knowledge. Building up institutional knowledge at your company can make you invaluable to everyday work.
- Work for small companies or on small teams. Smaller settings mean there are fewer opportunities for management and you will have to be a master of day-to-day activities.
- Mentor others. Mentoring is a great way to keep up with hands-on data work and coding even if you do go up the management ladder.
Management Isn’t the Only Way to Advance Your Career
Advancing in your career doesn’t always have to mean moving into management. For those who want to keep their hands in coding and data on a daily basis, Giri Tatavarty, vice president of data science at retail data and analytics company 84.51°, recommended investigating if your company has a technical track for career advancement. Those pursuing such an option can grow as individual contributors and basically become technical specialists.
“These experts typically spend more of their time actually developing the skill and techniques and experimenting, than managing people or working with the other other responsibilities of a typical management job,” Tatavarty said.
Tatavarty has pursued this technical track himself and is a senior individual contributor at 84.51°. He described his average day as being 30 percent to 40 percent working on innovative projects he leads and 20 percent to 30 percent on conducting reviews of projects being led by others on the team. The rest of the time is spent working on strategizing for the technical vision of the company, like determining the best strategy for scalable data science methods.
“Just like you need a top-notch brain surgeon to do your surgery or a very good engineer to build something really hard, you need a highly technical data scientist to solve business problems.”
“It’s a mix of strategy, innovation, day-to-day delivery and reviews of what is being done,” he said. While this is different from what he used to do as a more junior data scientist and has a broader scope, it all still involves daily coding.
“[It] would not be a tenable situation to be in the technical track and not code,” he said, adding that he would never want to. Answering business questions with data and math is a large part of what drew him to the field of data science, and the more he does it, the more questions he finds to solve.
More and more companies are starting to realize the value of offering a non-managerial career advancement track, he said. In his experience, companies have begun offering multiple advancement tracks over the past few years, especially at ones pursuing cutting-edge data science.
“Just like you need a top-notch brain surgeon to do your surgery or a very good engineer to build something really hard, you need a highly technical data scientist to solve business problems,” he said, giving self-driving cars and natural language processing as examples. “Many of the frontiers of data science require very specialized skill and experience and also focused time to spend on that.”
If a Technical Track Doesn’t Exist, Specialize
Even if more companies are recognizing the value of having a technical track for data science career advancement, that doesn’t mean every company offers it. If your company doesn’t, Tatavarty recommended pursuing the same ultimate goal of the technical track — to become a technical specialist — without the formal structure.
But that takes planning, he said, “because you only have so much time. These specializations usually take a lot more focus and going in a single direction, so there needs to be a plan to spend time and focus on one direction rather than trying 10 things.”
That plan starts with picking your domain or the area in which you want to specialize, he said. That could be technical areas like natural language processing, computer vision or A/B testing, according to Tatavarty, but it could also be business sectors like finance or advertising in which data science is applied. Creating a specialization plan also includes finding the right mentors and working with your manager to identify projects that will help you in your goal. Depending on the area of specialization you select, Tatavarty said it can take anywhere from a few months to a few years to become the go-to person at your company for that topic.
“Definitely, whether you have technical track or you do not have technical track, you will be highly valued and you will grow,” he said.
Stay Put Rather Than Change Companies
It’s hard to specialize if you don’t stay put for very long. There tends to be a lot of turnover in data science, with the average data scientist only staying at a company for 1.7 years, according to a 2021 report by 365 Data Science. This is in part because of the widespread opportunities in the field, according to Ramaa Nathan, director of data science at EVERSANA, a company that applies AI to rare diseases.
“The problem is, if you keep switching a lot, you do not build expertise in any one thing,” she said. Instead, she recommends that those who want to keep coding as they advance their career stick with a company. This not only builds up technical skills and seniority in the role, but also — most importantly — institutional knowledge within the company. That institutional knowledge is a specialization in its own right and will help keep you in a hands-on role.
Nathan gave her situation as an example. While she is a senior data scientist working with a team and does not do 100 percent of the coding like she used to in more junior roles, she guides and oversees the projects that others are implementing. Because of her institutional knowledge, she knows down to the smallest detail everything from the intricacies of client data to what her team has done on each project and is able to explain and correct mistakes directly. She described her position as director of data science as “a complete hands-on data science role.”
Within her own company, where Nathan has institutional knowledge and her direct involvement with coding and data is well-known, there is nothing preventing her from being hands on, she said. That would not necessarily be the case at another company.
“If I were to jump and go to another company, I can tell you that — if I’m expecting the same role — I cannot get hands on because that company might say, ‘No, at this role, we cannot expect you to be hands on and somebody else is doing it,’” she said. “Whereas by building up seniority in your own company, having that institutional knowledge, you can still be hands on in your higher roles.”
Seek Out Startups, Small Companies or Small Teams
Staying put to specialize won’t work for everyone. So, another way to advance your data science career without leaving coding or daily data work behind is to put yourself in a situation where you structurally can’t get away from the hands-on work of data science: Find a startup, a small company or a small team.
“They’re great opportunities for growth, but once you get to the top, you’re challenged to be a jack of all trades and master of all,” said Dylan Beal, vice president of analytics at Cane Bay Partners VI, a management consultancy. “Working with small teams means it won’t be easy to hand off all the data analytics and coding responsibilities. You get to be a senior contributor to the company while analyzing data, developing models and managing a small team.”
“Working with small teams means it won’t be easy to hand off all the data analytics and coding responsibilities. You get to be a senior contributor to the company while analyzing data, developing models and managing a small team.”
Elena Ivanova’s experience as the head of data science for CarParts.com, an online provider of aftermarket auto parts, bears out this dynamic. When she started with the company, there was a limited budget for data science and data analytics so they were not able to hire more data scientists for a while. This meant she had to be working with everything related to data at the company, which is exactly how she likes it.
“The data, for me, is everything,” she said.
Coming from academia, Ivanova was drawn to the field of data science by the allure of applying algorithms to real-world business needs and because she loved being an investigative researcher. Leading a small team allows her to do that every day. Even as the company has grown, the data science team has stayed relatively small. So, while she now oversees a team, she has not left coding behind.
Ivanova described her days as starting with checking in with her team and helping them work through any challenges they might have, but being mostly consumed with exploring data and solving problems. The cost and time related to international freight, and the impact frequent changes in those can have on the company, are examples of those problems she has to solve with data to help the company make good decisions that will keep it profitable.
What’s more, each new addition to the small team — rather than increasing Ivanova’s managerial tasks and taking away from hands-on work — actually increases her opportunity for coding and data work.
“Getting a new person allows me to bring more ideas that I can tackle and start working on,” she said. “Right now it’s a data science team that’s cross-functional, but still, we don’t have a lot of people so I still have a lot of opportunities. That’s why I’m still helping my people to look at data, look at the code [and] get some of the data modeling myself.”
Look to Mentoring to Stay Hands On
In some situations, management is an unavoidable element of becoming more senior as a data scientist. In those cases, technical mentoring of more junior data science professionals is a great way to keep your hands in coding and data.
Ashley Pitlyk, senior director of data science at Codility, a technical recruitment platform, described it as taking a “player-coach role.”
“Ensure that you’re still doing peer reviews of team code before models go into production,” she said. “Encourage your team to have development hours where they spend 1 to 2 hours a week learning something new and sharing it monthly or quarterly with the team — and take part in it with your team.”
Gong also recommended taking the player-coach role. It’s something he does with junior members of his team working on deadlines without tight timelines. This can allow him to keep his coding and direct data work skills sharp.
“It only takes a few months for you to be irrelevant if you stop coding.”
Mentoring isn’t just something that happens if you go down the management track either. Even though he pursued the technical track at his company, Tatavarty said that mentorship is a big part of what he does as a senior individual contributor. For him, technical mentoring happens when he is working through the science reviews of ongoing data science projects, but also when those leading the projects run into a challenge. Both situations require that he be hands on with the coding and data and keep very familiar with it.
But mentoring and staying hands on in data science have an interesting reciprocal relationship. Mentoring can help keep you doing coding and direct data work, but you also have to stay active to make your mentoring matter.
“It only takes a few months for you to be irrelevant if you stop coding,” Tatavarty said. “If you don’t code, what you’re saying is more opinion rather than substantiated claims from an expert or backed by data and you slowly lose your credibility or respect and people just listen to you because you are a higher pay grade. So you always need to be hands on.”
Don’t Be Quick to Dismiss the Traditional Management Track
While coding and direct data work is what draws many people to data science, and what many love about the role, Nathan urged other data scientists not to restrict themselves.
“It helps, in a way, to explore other opportunities and not just be stuck within the same thing,” she said. In her own career, her varied experiences — which started with coding out models for high-frequency trading through to project management and entrepreneurship — have made her more versatile in her current position as a senior data scientist. When a project manager on the data team left, for example, she was able to step in since she had experience doing project management.
Nathan also said that those other experiences were clarifying to her. She described her role as a project director as one where she was not able to work directly with the data.
“The few times when I did get access and I was able to code something, that was the happiest day of my life,” she said. That was her a-ha moment where she realized that, to be happy in her work, she had to be hands-on with data and coding.
“So, going out, doing something different, in a way makes you realize how much you really like it,” she said. “It helps to take some time to at least try other roles to know what it is that you really want.”
