Returning to School

"You'll never go back if you don't go now. It's too hard." I heard these words often four years ago, months before graduating from the University of Illinois at Urbana-Champaign. I had to decide between staying at Illinois for a Master's degree and accepting a job at a startup in Boston. I chose to work at the startup, because I wanted work experience and a change of pace. If graduate school was worth it to me, I trusted that I would come back to school.

Four years later, I missed research. Looking back at my time after graduating, I realized that I had clung to academic environments-- I lived near Harvard when I worked near Boston, kept up with my
research group at UIUC, and read various scientific and academic blogs. With encouragement from friends and family, I decided to apply to school. It wasn't easy-- I had a good life in Chicago, a good job, and loved the city. I just didn't see being fully satisfied with my future career without taking on more of a focus on research, and in order to get there I would have to return to school.

First, I had to choose between Master's and Ph.D. programs. I'm generalizing pretty badly here, but Master's degrees in Computer Science are mostly for people who want to continue working on software development, and focus on building systems that require a specialized body of knowledge. Ph.D. degrees are for those who want to focus on research. I chose the Ph.D. because I am more interested in research.

Applying to grad schools after working was not easy. I did not realize how important it was to keep in touch with professors in case I wanted to go back to school, and so I only had one professor to write a letter of recommendation for me. Fortunately, I had met a research scientist from Lincoln Lab while I was working at Firespout, and the CEO of my current company has a Ph.D. from MIT, so I assembled a good group of recommendations.

My GRE scores from four years ago were still valid, but I had never taken the Computer Science GRE Subject Test. I should have taken that earlier-- I would have done much better on it while the material was still fresh. I managed a good but not outstanding score there.

Crafting a personal statement was not easy. Do I focus on the research I did as an undergrad? If I do that, does it look like my time in industry was a waste? Do I try to paint my work experience as research-oriented? What do I want to research in grad school, anyway? No one told me that you won't be held to what you wrote about in your statement, so I thought I was writing a proposal for the next five years. After several rewrites and many revisions, I arrived at a statement I could live with.

Picking schools was difficult, too. It's hard to know how you'll be ranked in the field of this year's applicants. It's impossible to know exactly what life will be like at any particular school, unless you've attended it. I was fortunate to be accepted to a few schools that I was interested in, and by visiting each I found the best fit for me.

To any undergraduates who are torn between working and going to grad school, I do recommend working in industry for a few years. You'll develop good work habits, learn how to get things done in the real world, and learn what you really want to do with your life. Just make sure you keep in touch with professors for letters of recommendation (and their invaluable advice), and take whatever GREs you need before you graduate. Your test taking skills will be at their prime then.

Welcome

This weblog is about my graduate school experience. I start at Carnegie Mellon University in August, 2005, where I will pursue my Ph.D. at the Center For Automated Learning and Discovery (CALD), a department within the School of Computer Science (SCS). I am excited to return to school after working for four years in Boston and Chicago, and I hope to graduate in 2010.

At CMU, I will focus my research on Machine Learning, a field within Artificial Intelligence. It is the study of programs that learn from experience, with applications such as speech recognition, document categorization and summarization, and medical diagnosis. Arthur Samuel wrote one of the first Machine Learning programs, which learned to play checkers from a large volume of annotated games, and improved its play by playing against itself.

The CALD program is for students who are committed to Machine Learning research. This allows for a focused curriculum on topics such as learning algorithms and statistics. Students are not required to take classes on Operating Systems or Compilers, and there are no qualifying exams. Most other Ph.D. programs have more general requirements, which can delay the start of meaningful research.

I look forward to a challenging, productive, and enjoyable experience. My ultimate goal is undecided; I would be happy as a professor, an entrepreneur, or a researcher at IBM, Microsoft, or Google. I trust that I'll know the right path in a few years.