Transitioning from data scientist to data science leader

Stephanie Mari
Insight
Published in
9 min readMar 29, 2019

--

Insight Fellows in New York.

Over the last three years, I’ve worked with more than 500 Insight Fellows, coaching them as they transition to thriving industry careers in data science, data engineering, and artificial intelligence. Now, I’m setting my sights on helping Insight alumni continue to grow their careers, and one question that frequently comes up is, “How can I make the transition from an individual contributor to a leader?”

To get some answers, Insight’s NYC office hosted a discussion focused on the many paths of growth available to today’s data scientists. The panel was moderated by Kathy Copic, Insight’s VP of Growth, and featured several program alums working in NYC:

Below, I share three of the important lessons they’ve learned from their transitions from data scientists to data science leaders:

  • You will have less time for heads-down work, and that’s OK.
  • Your job is now about helping other people to do their jobs.
  • You can always improve at giving feedback.
L to R: Insight alums Deepna Devkar, Shanshan Ding, and Maureen Teyssier.

You will have less time for heads-down work, and that’s OK.

Though she was building a data product rather than conducting research, Maureen Teyssier’s first industry role at Axon Vibe (which uses location data to power the intelligence behind mobile apps) was similar to the work she did as an astrophysics postdoc: coding all day, focused on a single task, with little context switching. Her first experiences in management came two years later after moving to Enigma, a data management and intelligence firm. In her first six months there, she had no direct reports but worked very closely with nearly everyone at the company.

After leading her team on some shorter projects and establishing a track record of moving projects forward, she was soon in charge of core products and managing more than 10 employees. Her responsibilities expanded to include hiring to grow the team, holding regular one-on-ones with an increasing number of direct reports, coaching junior team members, coordinating with other divisions of the company, attending C-level meetings — and, when time permitted, completing her own technical work. Now, as the Director of Data Science and Data Engineering at Reonomy, a startup using data to transform the commercial real estate industry, Maureen is able to focus on leading her fast-growing team and driving strategic decisions at the company.

Coming from a PhD in Neuroscience, Deepna Devkar’s first industry role was at Viacom, where she worked on a team focused on asset optimization and audience research. This level of specialization provided an opportunity to dig deeply into both the business and data science side of problems, but at times she felt too far removed from the impact of her efforts. However, even as she enthusiastically interviewed for the role of VP and Head of Data Science at Dotdash — a smaller but growing digital media company where she would be able to work more closely with senior management—she feared that taking on a leadership role too early in her career would put her technical development on pause.

So, in her initial months on the job, she remained committed to keeping up with the cutting-edge data science techniques that were being used outside her company. She quickly learned that this feat wasn’t possible — nor was it the best use of her time. In a position like this, Deepna said, “your time is better spent learning business strategy” than diving deep into the hottest new algorithms. Today, she encourages would-be managers to remember that ‘technical work’ isn’t limited to coding. Her job requires her to switch between technical and non-technical collaboration throughout the day, constantly marrying the data science side with the business. She’s also still very close to her team’s projects, providing guidance on next steps and helping review their code.

In a position like this, “your time is better spent learning business strategy” than diving deep into the hottest new algorithms.

As she completed doctoral work in Mathematics, Shanshan Ding attended Insight to help transition from academic research to solving industry problems. Over her first four years in the industry — solving data science problems at YP Mobile Labs, Gilt Groupe, and then Compass — she was able to enjoy variety in her day-to-day tasks but didn’t always have a clear path for career growth. In late 2017, she joined the team at Hinge (“the dating app that’s designed to be deleted”) as a lead machine learning engineer and, after a year, was promoted to Director of Data Science.

As Shanshan learned, not all leaders have to totally give up individual contributor work. Because it is a smaller team, she is able to play a major role in decision-making at the company, but still has some capacity to complete technical projects. But, given her other responsibilities, there are real limits on the time she has available for deep, head-down coding. If a project requires a sustained focus and blocks of distraction-free work, she must delegate it to her team.

Your job is now about helping other people to do their jobs.

When you shift from being an individual contributor to a leader, your work is no longer just about you or what you can accomplish day-to-day. Deepna summarizes the role of data science leader like this: “your job is to evangelize data science and share what your team is doing to bring in more projects.” This requires you to immerse yourself into the business side of the problem, not just the technical side, so you have an accurate understanding of where your team can make the largest impact and how you can leverage their skill to achieve that.

Shanshan shared a similar perspective: as an individual contributor, your job is to advocate for yourself, but as a leader, your job is to advocate on behalf of data science. “Data science, while very much part of the public consciousness, is often a black box” to people outside the field; they may not be aware of everything it can do — or, conversely, may expect unrealistic results. “A manager has to advocate and educate” others at the company about what data science can and can’t (or shouldn’t) do.

“Your job is to evangelize data science and share what your team is doing to bring in more projects.”

While data scientists are often responsible for high-ROI work, they’re also quite expensive, so you need to know and be able to demonstrate how each member of your team adds value to the company, and increase that value over time. As Deepna says, “the onus is on you to elevate each team member” by helping to provide them with the experiences and resources they need to gain more skills. At Dotdash, as she has grown her team from three to ten people, she has also put serious effort into building out the company’s data science career ladder. It’s imperative that every employee has a clear picture of what the company needs from them, and what skills they’ll need to progress.

Whether or not the company has a formal career ladder for data scientists, as a manager, it’s still your job to help your employees grow. Maureen advises that a manager should freely share their reports’ accomplishments; this helps others see the value of the work, and opens the door for more of this great work to get done. Empower your team by not just explaining what to do or how to do it, but why it’s important to do. Challenge your team to stay aware of the fast-developing data science landscape and keep themselves up-to-date on the newest tools and tech stacks. But she also suggests providing some boundaries to help them make the most of what they learn: don’t “inappropriately build things” just to get a chance to play around with new tools. Use the tools that make sense to build what makes sense; make sure you’re always adding clear value, whether you’re using traditional or cutting-edge techniques.

Empower your team by not just explaining what to do or how to do it, but why it’s important to do.

As a manager, if your team is making an impact and helping drive success at the company, you may also be charged with helping to hire new employees. This was the case for Shanshan. She was the first data scientist at her company, and has since had the opportunity to start building out the team. She is taking a careful and controlled approach to growth — she doesn’t want to grow too quickly. If possible, her preference is to make one hire at a time so you can get to know them, ensure each person is properly and thoughtfully onboarded, and readjust your knowledge of the team’s cumulative skills and what they’re capable of. Then, you can decide if you need to add even more capacity, and if so, where to strategically add it.

You can always improve at giving feedback.

All of the panelists agreed that, whether you’re working with established employees or forming a relationship with new hires, some of the most important skills for a manager aren’t technical, they’re human. But most people are never trained on how exactly to be a manager, and there is a big learning curve for motivating and leading people. Looking back at the skills she thought she would need for a career in data science, Maureen reflected that “empathy is far more important than I ever realized.” From her perspective, empathy sets the framework for how to interact with people as people, not just employees. This is especially crucial when, in your role as manager, you have to hold a challenging conversation or deliver negative feedback. To carry out these conversations as productively as possible, she encourages managers to remember that there is no one-size-fits-all communication style. Some people will feel hurt unless you take a passive approach or provide an ‘enthusiasm sandwich’ that surrounds critique with praise. Others respond best to direct feedback and want to be explicitly told, “do not do this.” These are just two ends of the spectrum, and there are many points in between.

Shanshan — who joked that “you only need two skills in life: linear algebra and communication” — works on a team that practices radical candor, a feedback philosophy that says you can (and should) challenge people directly while simultaneously showing that you care about them personally. Even with this framework in place, it’s still challenging to develop a trusting manager-employee relationship where such communication is possible and feels safe. That connection can’t be built overnight, and it takes practice for managers to deliver tough feedback in a way that nurtures it.

For especially difficult conversations, like meeting with an employee about performance issues, Deepna sometimes works with Dotdash’s Human Resources team to practice what to say and how to say it. This is helpful because these conversations require both tact and clarity, and you have to be careful that what you say is what the employee hears. In a situation like this, you can’t use an ‘enthusiasm sandwich’ because it’s imperative that your message — there are negative consequences associated with continuing this behavior — are clear. When it comes to giving effective feedback, she says, “you never perfect it.” You just have to continue to learn.

At Insight, we work with the top companies, industry leaders, scientists, and engineers to shape the landscape of data. Insight Fellows don’t just go on to work in industry, they go on to lead industry. I’d like to thank our moderator, Kathy, and panelists Deepna, Shanshan, and Maureen for sharing some of the key lessons they’ve learned in transitioning from data scientists to data science leaders.

Do you want to transition to a career in data? Learn more about the Insight Fellows programs and start your application today.

Is your company growing its data team (including hiring for senior-level roles)? Insight can help! Learn more here.

--

--