Since it is the admission season, one often comes across excited students curious about various aspects of graduate life. Maybe, I will do a post sometime on ‘What factors must one consider before deciding which school to attend?’. For now, I will answer a popular question about an LTI program, which I have been often asked.

How straightforward is it to make the step up from CMU’s Masters in Language Technologies (MLT) program to the PhD in LTI program?

Update (Mar 3, 2020): The following answer doesn’t reflect the current scenario accurately. Interest in the fields of ML and NLP has grown significantly over the last few years, and I assume it is way more competitive these days than it was 4 years ago (when I joined the MLT program).

The number of people who continued for PhD after MLT at CMU are below. Likely, a few more were accepted and chose to go elsewhere.

2015: 832

2016: 1135

2017: 1131

2018: 1830

2019: 1344

Thanks to Robert Frederking, associate dean of graduate programs at CMU, for the statistics! Reference: here.


Originally Posted: I believe it is fairly straightforward. Unless it is a bad year for the department (limited funds, a larger MLT batch), close to 80%[1] MLT students are offered a PhD position after 2 years.

LTI has always adopted a model where they hire many Masters in Language Technologies (MLT) students, and offer only a very few direct PhD admits.

In most cases, this serves both parties well. MLT students can benefit from working closely with faculty members for two years. Further,

  • They can gauge their interests better. If computer science and languages do not excite you the most, there are various other great places to do whatever you like best. While most students continue for a PhD program at LTI, cases where students move to more relevant departments in CMU and other schools are not unheard of.

  • They can better assess if they really want a PhD. This time the decision is primarily driven by first hand experience doing research in a PhD like environment. A few students take up positions in industry (very often involving some amount of research, read research engineers). A few lean towards startups.

  • They end up with a better profile with established faculty members vouching for them the next time they apply for a PhD. Typically, students have an extra publication or two.

From the perspective of a faculty member, they get to assess student’s fit and value in their group very realistically in these 2 years. While most students end up getting a funded position with faculty members, the faculty members aren’t obligated to allocate funds for them during the MLT years. This is in contrast to the PhD students.

Hence, the process serves both the stakeholders well.

The only downside I fear is losing out on a few smart students who would rather take up guaranteed PhD positions in other schools instead of the MLT program.

[1] Fine Print: The number is more of an observation across the past two years, rather than an officially released figure.

-