Name: Alexander Lin
Class: 2019
Concentration: A.B. computer science / S.M. computational science & engineering
Hometown: Summit, N.J.
Internship focus: Data science
Internship location: TripAdvisor in Needham, Mass.
Describe your internship.
This summer, I was fortunate to join TripAdvisor’s Search Engine Marketing (SEM) team as a data science intern. My team is responsible for running paid advertisements on search engines such as Google and Bing with the ultimate goal of attracting users to our company’s website.
My primary project involved using machine learning methods to determine how much the company should bid on ad space, taking into account factors such as a user’s country of origin, travel destination of interest, and browsing device (e.g. desktop vs. mobile).
What is one of the most valuable lessons you learned from this internship, and why?
One important lesson that I took away from the internship is that effective predictive modeling in the real world is not always about using the fanciest model or implementing the newest research results. Sometimes, the simplest methods work best because they are intuitive, easy to debug, and quick to implement. Simply put, there can be several benefits to choosing linear regression over a neural network.
What is one of the biggest challenges you faced during this internship, and why? How did you overcome it?
Perhaps one of the biggest challenges was figuring out what data to use for my project. These tech companies record so much data to the point where it would be impossible to comb through everything without significant guidance. This made feature generation for machine learning algorithms quite difficult for a new member of the team, such as myself. Fortunately, I had many mentors whom I could consult for advice, making the data-collecting process much smoother than it would have been otherwise.
What skills from your courses at SEAS helped you the most during this internship, and why?
Courses at SEAS were instrumental in preparing me from this internship. Data science and machine learning classes walked me through the theory behind so many different modeling methods, thereby allowing me to know how to attack a variety of prediction, classification, and clustering problems. All of my courses were taught using Python and its associated libraries (such as Pandas, Sci-Kit Learn, and PyTorch), which have become the gold standard for data science in industrial applications.
Why has this internship been a good experience for you?
I enjoyed understanding how data science teams operate in an industrial setting. It was a very cool opportunity to work with real world data and develop actionable insights that could be built into production. It’s really cool knowing that your work has the potential to impact millions of users around the world.
How do you think this experience could inform or benefit your future career path?
It was very interesting to see how data scientists reduce business objectives into statistical problems that can be solved using their math/CS toolkit. I would definitely want to emulate their techniques for lateral thinking in my future career.
How did you find out about this internship?
I searched online for data science internship opportunities.