At the KDD 2019 IADSS Workshop, Amy Shi-Nash (Global Head of Analytics & Data Science, HSBC), Matt Curcio (VP of Data, Ripple) and Ying Li (Chief Scientist, Giving Tech Labs) discussed the current issues and future threats that may arise due to the lack of standards in the data science industry during the panel, moderated by Usama Fayyad (Co-Founder and CTO, OODA Health and Chairman, Open Insights).
The discussion started with the speakers’ thoughts on how a data scientist is different from a data analyst. According to Ying Li, a simple answer can be that a data scientist looks at the output, be it in the form of a code, insight, or a research paper. Amy was of the view that data scientists aren’t only knowledgeable but are also inherently curious, which explains why they ask so many questions and are keen on finding their own answers. Seeing fewer boundaries as constraints, they dig into the problem at hand, turn experimentation and learning mode on, and then find answers. Matt highlighted that another difference lies in the main tool: a data scientist uses Python while an analyst relies more on SQL. At the baseline, data scientist professionals and data analysts are involved in researching, and thereby, an analyst position can be a great spot for data science candidates.
At the baseline, data scientist professionals and data analysts are involved in researching, and thereby, an analyst position can be a great spot for data science candidates.
As the discussion continued, the speakers agreed that in an organization, the roles of data science experts, machine learning engineers, and algorithm developers might seem a bit similar. Amy emphasized that it’s crucial for mid-level management to understand the different types of scientists working in the company and the specific roles they play. Realizing what a data scientist is capable of can help with business sharing and collaborations. Regardless of the specific skill sets possessed by these professionals, it’s important that they work together as a team for a shared goal. Hence, a career model that specifies particular roles of professionals can make it easy for the team to gel up.
Ying further commented that depending on the type of company they work in, data scientists may focus on dealing with the business and/or clients to clarify the problem and find an ideal solution while using their expertise in research, algorithm, and coding. On the other hand, a machine learning engineer may spend years doing classification only. As a data scientist, you’re likely to focus primarily on algorithms whereas; a machine learning engineer goes a step ahead and masters the art of algorithms as well as polishing their engineering skills. While they’re more in demand, transitioning from the role of a data scientist to a machine learning engineer may not be so easy.
It’s been coming up a lot lately that on an organizational level, there are no global standards to identify data science roles. Amy stressed that it’s because the industries across the world are so diverse. This point brought up to the main topic: candidates’ assessment and matching the right person to the job. Sharing his thoughts, Ying Li mentioned that a precise assessment of an individual’s capability and skills is essential when it comes to hiring data scientists. The core focus should ideally be on coding, insight, and algorithms or research. Secondly, the interview questions should be such that they give a bird’s eye view of the candidate’s skill set and help assess their capabilities, in theory, coding, critical thinking, math, and algorithms. The bottom line is to focus more on testing the candidates in high pressure situations while asking simple questions than on their resume.
On an organizational level, there are no global standards to identify data science roles since the industries across the world are so diverse.
In reply, a listener in the panel expressed that not all individuals who fail to perform well in high pressure situations are bad candidates. The speakers acknowledged that the pressure-inducing technique helps to filter the ones who are capable of thinking and acting fast. It’s ideal for startups that are looking to hire the first data scientist for their company. On the contrary, if a business wants to hire a professional to work on long-term research projects, they might not be interested in assessing how fast they can work. At the end of the day, it comes down to the specific requirements of the company running the recruitment campaign.
The discussion then took a turn when Usama asked for thoughts on the future of a data professional. Amy accepted that technological advancements had created a more diverse and challenging work-space for data professionals in the modern world. She implied that they must identify their key strengths, particularly related to critical thinking and data interrogation. Another way to keep with the fast-paced industry is by learning things quickly.
Going forward, the trend of people belonging to different fields switching their career paths to data science was discussed. Amy suggested that these candidates can bring a lot of diversity in the team, and so, managers should nurture this trend. For people who’re just starting in the data science field, she advised that learning to use the basic tools is useful, but the deeper you dive into a specific niche, the more you limit your abilities. It depends on what your interests are and which direction you want to steer your career in.
An audience member then highlighted that managers and recruitment officers are often interested in testing the analytical mindset of the candidates and inquired how they do so. Ying Li volunteered to answer and said that one way of doing this is by giving them a business problem to solve and see how they think through it. It also works if you give the candidates a chance to teach you something by explaining their thought process when they tried solving the problem. Furthermore, communication skills are of utmost importance in the data science field. Employers may utilize the storytelling approach, asking the candidates to narrate a problem and the outcomes. It’s a simple way to see if they like to make things simple or to complicate them.
An important concern in the field of data science is that it allows you to grow for a couple of years, and after that you’re just stuck in one place. It is the reason for a noticeably high attrition rate.
Another important concern in the field of data science was then discussed; how it allows you to grow for a couple of years or so, and after that, you’re just stuck in one place. Matt believed that it was the reason for a noticeably high attrition rate. He suggested that the best way to deal with this is by making sure there’s a clear, exciting growth path for data scientists in a company. The employees should be able to talk to their managers about their growth rate and opportunities. In essence, data science is a field that offers a lot of variety in what you do. There’s so much to learn and experiment with – the process shouldn’t be repetitive. Transitioning from a small team to a larger one that handles different tasks shouldn’t be difficult if the team is agile and well-integrated.
"The industry is evolving, and with automation, data scientists can expect to see some tasks being pushed down to less technical people. However, the challenging nature of the job won’t change. No matter how much the technological world advances, machines can never learn the art of critical thinking."
Towards the end, Amy talked about how a regulatory environment may stop data scientists from using some modern tools that they think they should use. Even in such a case, she believes that the data science job doesn’t have to be boring. You can, and you must follow the rules while working on modeling parts, but it doesn’t mean you can’t get creative with data interpretation. Matt commented that the industry is evolving, and with automation, data scientists can expect to see some tasks being pushed down to less technical people. However, the challenging nature of the job won’t change. No matter how much the technological world advances, machines can never learn the art of critical thinking.
Comments