Erin Boehmer

Erin Boehmer graduated from the School of Information in 2015 with a master’s in data science. Prior to the I School, she graduated from the University of Virginia with a bachelor’s in systems engineering and computer science. Currently, she is a staff machine learning and data platform engineer at Carta and Equilibrium Energy.

About two months after completing my MIDS degree, I packed up and moved from Boston to Kampala, Uganda, to accept a data scientist position with Fenix International (now ENGIE Energy Access). The company has developed a pay-as-you-go solar kit that collects amazingly rich payments data for people in rural Uganda who are completely off the grid and therefore generate little to no data footprint otherwise. I [was] using this data set to understand customer behavior and inform business strategy so that, as a company, we can more effectively replace dangerous light sources, like kerosene lamps, with safer, greener, and more affordable solar.

I rely heavily on the knowledge and confidence that I gained from the MIDS curriculum as I approach[ed] open-ended data science problems to inform company strategy at Fenix. As the sole data scientist in the company, I [was] often approached with questions that have no clear answer; my professors at UC Berkeley, however, have provided me with a strong basis in structured problem solving and critical thinking as a data scientist. I am often thinking back to our discussions on the importance of testing assumptions for various statistical methods, evaluating more than just the accuracy of a model, and considering the ethical implications of the models we are building. As I have learned from my Fenix colleagues, these were the skills that so many applicants to the data scientist position were lacking. MIDS graduates have more than just the skills to “do”; more importantly, they have a steadfast curiosity to understand.

My favorite course in the program had to be “Machine Learning.” I loved seeing the power of statistical modeling for practical prediction and data-driven strategy. From clustering instances of the Iris data set to predict the flower type of new data points, to predicting demand for the Capital BikeShare program in Washington, D.C., each IPython notebook I worked on felt like a playground for statistical inquiry and strategic understanding.

I love data science. I had so much fun studying in the MIDS program, constantly challenging myself to learn more and think bigger. The problems we work to solve as data scientists are difficult, and can at times seem intractable, but knowing that our hard work will lead to valuable insights and significant impact fill the problem solving process with purpose and excitement. This is a career path that can grow with you and will never get old.

Don’t underestimate the importance of the ethics course. Coming from an engineering background, I know there is a tendency for technically minded students to treat these subjects as “soft” and therefore less deserving of attention than, say, the “Machine Learning” or “Scaling for Big Data” courses. Yet I’ve found that in my current position, I constantly need to evaluate how my models are going to affect every different customer profile. While Fenix is a for-profit venture, we also exist to serve people at the lowest baselines who, one day, could stop paying for their solar kit because a family member has become critically ill or their home burns down. We want our models to distinguish between the good-intentioned and the bad-intentioned and allow for recovery with good behavior, especially given the lack of exposure to ideas like credit scores or micro-loans. The things you learn in the ethics course may seem obvious when talking about cases like the Milgram Experiment, but the line between “ethical” and “unethical” is much more controversial in models that are cutting edge.