Today we had the pleasure of having Sudheer Marsetti speak to us at Metis. Currently a Senior Data Science Director at Aetna, Shdheer was very open in sharing about the data science group that he manages. Although he started off by admitting that he didn’t have a fancy presentation or talk prepared today, he spoke candidly. By the end of the session, no one got enough of his incredible data science insights, advice, and past projects that he called “war stories”.
As with all our guest speakers, we first ask Sudheer what his best data science resource is and if he has any advice for new data scientists. He shared that his online subscription to Safari Books with books and videos very useful. At work, he’s using them for reference and at home he’s using them to learn more.
His advice for students: working just 8 hours a day will not make you a data scientists. He advises us to work hard - to the point where work-life balance means there’s no differentiation between the 2. Before he started, he also encouraged the class that he looks for talented data scientists at bootcamps.
Sudheer started our his talk by walking us through the pipeline and data science team’s structure at Aetna. For a company the size of Aetna, naturally there are a lot of moving pieces so he has defined 3 main roles: Data engineering, data science, and data visualization.
On his team data engineers, who are responsible of the initial data gathering and munging, actually does 80% of the work. In fact, there’s a 10-to-1 ratio between data engineers and data scientists. And with the rising importance of visual analytics, Aetna has a dedicated team that does data visualization.
Sudheer also introduced the role of the Data Science Engineer who is adept at both computer science and data science. While data scientists are great at math and statistics, they tend to write lousy code. But a data science engineer, knowledgable in machine learning and programming, can take the complete model and write end-to-end code with Python, Java, or natively in Spark with Scala.
What does he look for in a data scientist? Someone that’s smart and analytical, because he can provide technical training in machine learning. But not soft skills. So in addition to being able to pass a technical coding interview, it is very important to be able to work hard and work well.
I want to thank Sudheer for coming out to Metis, and for speaking to us about data science from a much needed management and business perspective. You can find Sudheer Marisetti on Twitter @abacus_concepts-->