GHC 19 -Demystifying Careers in Data Science

This is session notes for a GHC19 Session — “ Demystifying Careers in Data Science” held on 4th October, 2019.

Speakers:

Anna Coenen — Senior Data Scientist, The New York Times

Stephanie Yang — Senior Staff Data Scientist, Foursquare

Tamar Shapiro — Head of Analytics, Instagram

Geetu Ambwani — Data Insights Lead, Flatiron Health

Anna began the session by asking panelists to introduce themselves and asked them what their most favorite and least favorite part of their jobs were. Stephanie said that she loves that her job allowed her to wear many different hats and that she got to do a lot of software development. She said her company has a brand management problem in that not many people know what they do. She said that this makes hiring difficult. Tamar said her favorite part was working with data to help solve problems. She said a hard part of her job, but not one she hated was making hard decisions. Geetu introduced Flatiron and gave an overview of how they build cancer care products. She said that she loved working in one of the first product based health tech companies that looks at actual patient data to solve problems. Anna said that she loves working on recommendation systems and seeing user insights.

Anna then asked the panelists to talk about how they ended up in their current career positions and what titles they had previously held. Tamar said that she started her career as an actuary. She said that a lot of her job is product analytics and explained how this helps drive product strategy. Geetu said that she pursued Computer Science in graduate school. She was working on speech and natural language processing in school, but there was no official Machine Learning Course. She held the title “Senior Research Scientist” for almost three jobs. Stephanie worked as a mathematician faculty, then in finance as a quantitative developer. She now works in data science, but is actually a Software Developer by title. She said that she believes “Causal Inference” and mathematical modeling are great branches of data science. She also said that Machine Learning is a tiny bit of the quantitative field. She likes to be known as a quantitative developer.

The next question Anna posed was if there are some skills that are a common requirements among the different roles that come under the data science umbrella. Geetu said that the best thing a data scientist does is taking an ambiguous problem and transforming it into an analytical problem. Stephanie said that she did not believe that there is a single commonality between all the role requirements. She said that data scientists do something in common, but that is not limited to them. Other branches of tech like software, management etc. also have these qualities.Tamar said that these differences are the reason why it is important to describe what different teams do. All the jobs the panelists did were entirely different, but they were all data scientists. The field has a wide range of jobs.

Anna then enquired how data science sit within the organization of the different panelists. Geetu said that there is no data science team in her organization. They have software, data insights, business and other departments. There is clearly an overlap between them. Geetu said that they have stopped defining themselves based on skills and talk based on what kind of job their departments/teams do. Geetu said her team was called data driven product discovery. Tamar said that Data Science is part of product engineering in Instagram. She said that there are designers, product managers, engineers who are all part of this team. Data and analytics is a core part of entire ecosystem at Instagram.

Anna then asked the panelists the much awaited question of how to get into data science. Stephanie said that panelists had all grown their careers in the quantitative field. She referred to an article “Data science is different now” by Vicki Boykis. She said that it was alarming that the number of entry level positions for data science are less than the number of graduates. She said that the blog had an interesting piece of advice to get into an organization with some other role and then switch into a data science role. Stephanie said she does not completely agree with this. Stephanie also said that if you work in a smaller company, you can wear different hats and figure out what to do. Geetu said Data science is increasingly becoming specialized in companies that are technology first. But in industries like healthcare and media, data science is not yet considered to be a first class citizen. She advised that growth of data scientists is better in these industries.

Anna followed this with a question of whether graduate level programs and bootcamps are even useful. Tamar said that most data science job requirements have experience listed in them. Stephanie said that if one can get into a good reputed program, they should go, as it would give good job placements.

Anna concluded by asking the panelists what data scientists could do to advocate their field within their organizations to make their work impactful. Geetu said that there is a difference between being a first data scientist in a team and joining an existing team. She said as a first data scientist, you should not get hung up in the role and be ready to do any job presented to you. Tamar said as long as you and your team have the same shared goals, you can use your skills. As a data scientist, you are working with others towards the same goal.

Anna then opened the sessions for questions from the panelists.

Q1. I want to get into data science. How do I get back into the math side of things that I missed at college?

Geetu said that Coursera and Data camp are good resources to learn from.

Q2. As an undergrad, what kind of things do you think I should learn to become a data scientist?

Stephanie replied that coding and math classes are important.

Q3. What are key qualities required to progress your career in data science?

Tamar said autonomy is a key requirement — to be able to identify problems that can be solved with data. Communicating to your audience who might or might not be technical is also essential.

Q4. What is the absolute baseline for the job?

Stephanie replied that Exploratory Data Analysis (EDA), 1 course at least in each of coding, statistics and Machine Learning are useful. Geetu said that being familiar with an analytical language like Python and R and having used it to analyze data is key. She also said that SQL is important.