Learning Data Science Outside of the Classroom

As a student pursuing my Master’s degree in Electrical Engineering at the University of Pennsylvania, I’m excited to have the opportunity to put into practice what I am learning in school through a data science internship at Punchh.

It’s not often that data science is associated with electrical engineering, as it is to computer science. The Electrical Engineering program at Penn is incredible and provides a lot of freedom. It gives me the chance to pursue my interest in data science and help sharpen my skills. Tough optimization problems, including efficient data transfer and sensor placements, are being solved in the electrical engineering department. Optimization is the core of data science. That is why the Penn Electrical Engineering program, with its flexibility and excellent faculty focusing on optimization, is the path I chose.

The projects I work on at Punchh directly relate to classes I’ve taken at Penn. Courses such as machine learning and statistics gave me a basic understanding of machine learning concepts. A course in data mining showed me how to apply the concepts on large data and scale the algorithms to accommodate huge amounts of data. Along with these, my course projects on sequential data analysis came in handy at Punchh, as I’m working with text data.

Data science at Punchh is unique because I get to build things from scratch. The data science team recently developed, and it’s exciting to contribute in the early stages. I get to work on exciting deep learning models and read about the latest advances in tech as part of my job. I have the opportunity to learn a lot at Punchh, such as working with a team, managing code, and meeting deadlines.

In my internship, I’m currently working on a deep sentiment analysis project. It’s an engaging Natural Language Processing (NLP) project where I use various deep learning frameworks to classify user reviews into the sentiments they express. It’s always exciting to see how a machine would do when compared to a human. It’s also interesting to figure out ways to improve the algorithms, so they can get as close to human accuracy as possible. My current goal is to improve the accuracy of the deep sentiment model and to optimize the algorithms to work faster.

NLP is an incredible field. There is a limitless supply of words. We as humans, with generations of knowledge, struggle to speak more than two languages. Yet we build algorithms to transcribe, segment, and classify language. This unbelievably challenging task is what’s interesting. It’s fascinating to think, “how can we make a machine achieve something that took us evolutions to get to and how can we do it with the limited resources we have!?”

It has been great to have the opportunity to work with a team at Punchh. In college, each project has its requirements, which are open-ended and defined by me. At a company, the objectives of the project align with the company’s vision. Being part of a team, I get to learn how to tackle problems with my coworkers and exchange knowledge with vastly intelligent people.

With my degree, I want to gain more knowledge in data science and continue to work on thought-provoking projects. After I graduate, I would like to work on real-world problems in a professional setting, like I am currently doing at Punchh.

Chetan Tutika was a Data Science Intern at Punchh who focused on Natural Language Processing and deep learning models which deal with interpreting text. He is currently pursuing a Masters in Electrical Engineering at the University of Pennsylvania. Prior, he graduated from V.I.T in India with a Bachelors in Electronics and Communication. In his free time, Chetan likes to play guitar and is part of an Indie fusion band called Penn Sargam.