Hangar Team Spotlight - Anshu Sinha on the Data Science Career Path

Hangar Team Spotlight:

Anshu Sinha on the Data Science Career Path

At the center of Hangar’s work and the work of each of its portfolio companies is data science, a field that’s rapidly expanding and evolving across every industry. We spoke to Hangar’s Chief Data Science Officer, Anshu Sinha, about her data science career, the importance of data science in the public sector and a new matrix she developed to plot out what a career in this growing industry looks like.

How did you get started in data science and what brought you to Hangar?

I found my way to data science after getting my PhD in computational biology, where I first got into statistics and machine learning. At the time, data science didn’t exist as a field but I had a few friends who were early data scientists -- it sounded perfect for me so I set myself on that path. I started and grew out the data science team at ZocDoc. While I was there, I really started to think about how data teams should be organized and how they could be most productive. 

Eventually, I ended up at Hawkfish to develop and analyze voter data around the Presidential election. Improving voter data is a challenge within the democratic ecosystem -- we need to engage in more consistent efforts towards improvement. We used our voter data to help with turnout efforts such as registration, mobilization and early voting, as well as persuasion on candidates and issues. It’s a really interesting machine learning challenge -- we needed to find both high-quality and expansive data and put it all together to develop a really clear picture of a voter and what moves them. 

Now at Hangar, I am the company’s first Chief Data Science Officer. I was drawn to Hangar because I felt that Hangar and its portfolio companies’ mission-driven, data-focused approaches were a perfect fit for my interests. Our portfolio companies are all early stage and need a guiding hand to help build them out. That’s where I come in. My career has been focused on building and structuring data science teams.

How can data science be better used in the public sector?

It’s hard to imagine a startup or other entity in the tech world that isn’t at least thinking about analytics and data science. But, the government hasn’t caught up. I saw this when I started working at Hawkfish -- there’s no agency overseeing data science, there’s no data science role in our digital service. There are studies showing that data science can improve a business’s bottom line by as much as 20 percent because it improves efficiency and, even beyond that, it opens up new lines of work. So, there’s a huge opportunity being lost in the public sector.

The obvious question is, why? First, sectors of the government have been slow to adopt digital technologies overall, so this is simply an extension of that. But also, I think there is generally a lack of understanding of what data science is and what it can do. Moreover, there can be a fear or mistrust of data and its analysis, especially if it’s not telling us what we expect to hear -- we all know that data can be misinterpreted at best, manipulated at worst. 

Probably the most common application of data science is as predictive analytics used to improve outcomes -- using historical data to predict future outcomes which you can then plan and strategize around. Good data science requires good data, significant domain knowledge and an advanced set of interdisciplinary skills including machine learning, statistics and computer science to be successful within the context of a business. In other words, it requires a real investment of resources before you can achieve results. And once you achieve results, it takes even longer to confidently prove success against a baseline. In our line of work, we’re using data science to improve efficiencies and existing processes, all of which takes time to be proven. Contrast this with other product improvements where a gain of functionality or an enhancement is clear and deterministic -- a new feature, a user interface change, etc. 

Because data science can add such value, however, it’s so important that it gets a seat at the table. What’s so unique about our work at Hangar is that each of our portfolio companies is tackling a slightly different public sector problem -- whether wildfires or healthcare and so on -- and applying data science to each problem in a new way in order to provide better solutions and outcomes for people.

You worked to develop a data science career ladder (see here). Why is defining the data science career path so important?

Data science as both a field and a practice is still evolving. So, for those navigating a career in data science, the path is murky, to say the least.  For myself, I didn’t have a career path to refer to -- I just thought that you worked hard and got rewarded. A lot of my motivation in creating this matrix was to show my team what this career can look like, but also, I wanted to give an objective and consistent rubric for evaluation and promotion. I’ve found that laying this out with an employee from the start is invaluable in setting expectations early on. 

While there are many career ladders out there for tech roles in general, there are only a few for data science and they have only just started coming out over the last year. I hope that this can also serve as a resource for the general community, leaders, and new data scientists. I’ve been working with it and developing it over the last five years, so it’s lived out in the wild for some time now and I wish I had published it earlier!

Can you take us through the matrix?

This matrix breaks down role definition into three key areas: getting things done, expertise and teamwork and communication. There are probably a whole host of other things we could put in there, but based on my work experience, I felt that these three covered a lot of the ground that is important in being successful as a data scientist. Within these sections, I broke down getting things done to execution quality, deliverable types and pace; expertise to core competency and field/industry knowledge; and teamwork and communication to its namesake attributes. 

I covered five career levels here: Associate, Mid-level, Senior, Principal, Director and VP. It’s worth noting here that many other career path ladders break out individual contributor (IC) paths from managerial paths. Here, I can imagine the Principal level being fleshed out further for increased advancement as an IC. But beyond the Principal level, I focused on the managerial roles. To be quite honest, I wanted to lay out my goals as a manager and my expectations for promotion to my (non-technical) bosses -- sometimes you have to be proactive!  

The matrix goes into a lot of detail on the skill set at each level for each criteria. At a glance, entry and mid-level is all about learning, jumping in and being a great team player to build out all of your skills. Responsibilities and purviews increase at the senior/principal level and expectations become higher. They own and manage projects of increasing complexity and are expected to execute against set KPIs. They help develop the team and grow as thought leaders across the organization. At the most senior and more managerial levels, we layer in management, corporate strategy and organizational goals and how data science aligns with that more broadly. As we get more senior, data science career progression really just parallels many other career paths. We gain experience, we take on more responsibilities and increase our purview, and this defines our value. The goal of this matrix is to give detail on what this path looks like for data science. You can find the matrix here.