Graduate School of Arts and Sciences

February 24, 2015

Jackson Van Amburg | Graduate School of Arts and Sciences

Brandeis GSAS: What was your education background prior to coming to Brandeis for your Masters? Did you have existing knowledge in Computational Linguistics (CL)?

Keelan Armstrong: I went to Indiana University and majored in Theoretical Linguistics and Arabic. I didn’t really know CL was a thing until my junior year, which is a shame because Indiana actually has some amazing people there doing CL. Anyway, I took one programming class and one CL class my senior year and realized CL could provide an avenue of linguistics that I had never even imagined.

Theoretical linguistics can be pretty dry, or at least I felt the “theoretical” aspect a lot my senior year and wanted something more hands on, which is what led me to CL in the first place. So I happened to apply to Brandeis at the last second for their CL program. It was kind of a leap of faith because I didn’t know if I would be able to keep up with CL or would even continue to enjoy it as I delved deeper into the subject.

GSAS: Were there any other schools you considered besides Brandeis?

KA: Brandeis was the only CL program I applied to. I applied to Indiana University (IU) for more theoretical linguistics, also IU for Arabic language and literature. I also applied to the University of Washington (UW) for their Arabic language and literature PhD and I think Wisconsin for biblical studies. I was very unsure about what I wanted to do and wanted a lot of options.

I got accepted to everything except the Biblical studies program. I visited UW and Brandeis. At the end of the day though it was obvious that Brandeis’ CL program at least had a lot of funding - I think James Pustejovsky said they have some $2 million in funding.

I also got the impression that there was a strong emphasis on placing people into either the industry or academia and at that time I was still considering getting a PhD someday (though I still might someday), and I knew I would be able to either go straight into a job or keep on with academia relatively easily, depending how I felt at the end of the program.

GSAS: That’s quite the scattershot of schools and fields…and the potential for placement in either academia and/or the job market is certainly important, especially with the weight of student loans after graduation. You mentioned a “leap of faith” with CL, specifically related to concern over whether you would continue to enjoy it the deeper you immersed yourself in it. What was (assuming there was one) the point where you realized this was the right choice for you? When did it stop being a leap of faith and you started to enjoy the fall, so to speak?

KA: I realized it was the right choice for me probably half way through the first semester. The program does a good job of propelling you right into the heart of all the areas of CL, while at the same time still having quite a bit of general programming involved. I came to enjoy programming languages just as much as natural language and it was amazing at how much CL and basic programming skills made it possible to let me revisit old problems in theoretical linguistics and tackle them from a seemingly completely different angle.

GSAS: Was there any particular class that stands out as being particularly valuable, either at the time or in retrospect?

KA: Bert Xue’s Programming for Linguistics (Ling 131) was a really good class. Lots of live coding by him and I think he does an amazing job of taking people from almost no programming experience and works up to writing pretty large amounts of code - though part of that should just be credited to the Natural Language Toolkit (NLTK) textbook, which is free online. That’s the class that really let me feel comfortable programming in Python and through working with NLTK, I got access to a lot of interesting tools (like machine learning classifiers!) without having to necessarily be able to implement them myself.

Sophia Malamud’s “Mathematical Methods in Linguistics” class was also really useful, but not until two semesters later when we were actually implementing machine learning algorithms.

GSAS: Along those lines, Brandeis is one of the only schools that offers a two year Masters program in CL; most others are one year only, with the trade off being less expense incurred on the part of the student in the long run. Now that you are in the job market on the other end of this experience, do you feel it was worth it? Do you feel more prepared/capable and competitive as a result of the extra time?

KA: Oh, that’s interesting. I didn’t know that most other CL programs are one year, but I think Brandeis’ program is also different in that it doesn’t require as much programming background, though that’s changing too. Anyway, I wish I could’ve stayed there longer and taken more classes actually. There are a lot of topics, both in CL and general computer science, that I could study for years and still not be half as good as a lot of people in the industry.

I was pretty new to programming too, and I think by the end of the two years I had been exposed to enough programming and quality code that I could feel comfortable writing quality code myself, and that’s probably the most important thing that companies test applicants on.

GSAS: Another concern for any student investing in graduate education is the job search/placement process near the end of the degree. Did your job search start while you were still working on your degree at Brandeis? Was there any fear of not having post-masters placement somewhere and were there any connections through Brandeis that proved helpful?

KA: So the job search started while I was at Brandeis, which provides an amazing amount of support in terms of networking and career fairs. They have connections not only with smaller local companies and start-ups, but also a lot of big name companies, like Amazon and Apple. A number of professors there, such as James Pustejovsky and Marie Meteer, are also actively involved in the industry and in my time never failed to be helpful whether it was in terms of my resumé or advice on the interview process. James actually connected me with the main developer at Vioby which led to my internship while I was in school. I was pretty busy my last semester with phone interviews and ended up with multiple job offers, which I could not have done without Brandeis’ extended network of people in the industry. However, I found Narrative Science kind of randomly though, through Indeed.com.

GSAS: So what is Narrative Science and what do you do there?

KA: Narrative Science is a company that takes other companies’ data, does some black magic to it and generates natural language reports. We do everything from generating recaps from baseball to creating stock reports, I think they’re officially called attribution reports, for hedge funds. Our goal is to be able to take raw data and extract the interesting and most importantly useful information and put that in the hands of people who can ultimately make the most informed decisions possible.

GSAS: So where do your CL skills come into play? It sounds like they employ a lot of people who double majored in English and statistics. Are you the “black magic” component?

KA: I probably add some “black magic” though there are plenty of people there who are very knowledgeable about certain areas of CL already - statistics will get you very far in most CL concepts. I currently work on the Data Team there, so I prepare data and work with people who have domain specific knowledge to be able to create natural sounding reports. I’m about to start working on baseball data actually, but I first have to sit with someone and learn all the rules and jargon associated with it. Anyway, it’s really interesting from a CL point of view, because most companies are trying to extract structured data from text, not build free text from structured data. Finally though, we’re hiring if any Brandeisians are looking to move to Chicago!

GSAS: Is there any advice you’d give someone looking to take the “leap of faith” like you did? Any suggested reading or things online they should look at if they are on the fence about the field or want to learn more?

KA: My only advice is to look at the employment rate for graduates in CL or computer science in general. It’s a really good time to be doing this type of thing; I feel like we’re just on the tip of the iceberg in terms of what CL will be capable of and there is a lot of interesting research going on at Brandeis. My original belief that I would have both the doors of academia and industry open to me proved to be correct, and in my job I still have the opportunity to do research or learn about new CL topics, and there are definitely a lot of jobs in this field that heavily focus on research too.

I should probably touch on just how much support Brandeis was able to offer. IU was a big state school and there were times when I felt more like a cog in some great machinery. Brandeis had a much more personal feel and a lot of CL classes are able to tailor the topic depending on what people are interested in.

GSAS: Ok, last question - if you were talking to yourself before Brandeis, what would you have wanted to hear/know?

KA: If I could talk to myself…I would just say that I got an awesome job!