Search
Close this search box.

Alumnus André Martins – A Path Driven by Industry-University Collaboration

Alumnus André Martins, now working at Priberam, talks about his experience so far: A Path Driven by Industry-University Collaboration

Andre Martins The 5th Lisbon Machine Learning School starts today, in Lisbon, and finishes on July 23th, 2015. This is a key event in the Machine Learning field in the world, and one of its mentors is André Martins, our alumnus.

André Martins was the first graduate of the Language Technologies dual degree program at Instituto Superior Técnico (IST) and Carnegie Mellon University (CMU), in the scope of the CMU Portugal Program. During his studies, he was distinguished with a Best Paper Award in 2009, and the IBM Scientific Prize in 2011, having later interned at Google. When so many young researchers of his generation leave Portugal to look into for better opportunities, André Martins has chosen to stay in Portugal, and to return to the company where his career started: Priberam.

André had already been working with Priberam when he enrolled in the dual degree doctoral program, in 2007. After defending his dissertation, in May 2012, he returned with new ideas from that experience that included two years at CMU and an internship at Google for four months. Priberam Labs was one of the initiatives he spearheaded at the company, as a “bridge between industry and academia”, with the goal of putting into practice the best ideas in the field and creating new technology. It functions as an “innovation probe” that is now focusing on European projects.
One of the largest initiatives of Priberam Labs, mentored and stimulated by André Martins, is the Lisbon Machine Learning School, a week long summer school that is growing steadily in recognition and industry support. According to André Martins, it “fulfils a space in Machine Learning.” Its main purpose is to show students the latest research in the field, and have them share their work with each other; but mostly, to network with teachers, researchers and industry. This event is hosted by IST and the Spoken Language Systems Lab (L2F) of INESC-ID and sponsored by Google (that provides attendance scholarships), Priberam, Feedzai and Unbabel. André Martins credits the success of the summer school to the “core team” of speakers, like Noah Smith, from CMU, and Slav Petrov, from Google, that “consolidated the school.”

Another idea André Martins brought to Lisbon, inspired by his time at Carnegie Mellon during his PhD studies, are the Priberam Machine Learning Lunch Seminars, held every two weeks at IST. The objective is to “develop critical mass around certain topics” and provide the students with an opportunity to showcase their work and get comments on the work they are developing from peers, fostering initiative and autonomy. In their sixth academic year, the seminars have featured CMU faculty and CMU Portugal doctoral students, as well as industry people.

Keeping a close connection to academia, in particular his Portuguese alma mater, IST, is important to André Martins, where he “tries to contribute so that some of the things (practices) there can be done here.” Beyond the seminars, he advises doctoral students, including our dual degree Ph.D. student Zita Marinho. He also remains connected with his advisors, both at IST and CMU, by developing papers and overseeing Ph.D, students with João Paulo Costeira, and developing a new project with Noah Smith.

The time spent in Pittsburgh had a strong impact on André Martins, a “vibrant environment” from which he maintains many ties. The differences between the academic and corporate cultures of Portugal and the U.S. are what inspires him, “CMU has a strong connection to industry (…) a concept of doing things with a practical use.” He calls it “industrial research” which is an idea also very strong at Google, “they are obsessed with efficiency and the impact of their products.”

The Machine Learning and Computational Linguistics field is still a young one, especially in Portugal, and its research community is sparse. Research groups are emerging all over Europe in the recent years, but the multidisciplinary nature of the discipline means that are yet few graduates in Machine Learning itself. The CMU Portugal partnership’s Language Technologies program has had eight students since 2007, of which two have already graduated. André Martins was the first student accepted in this program, and the first student to graduate from it.

When asked why he stayed in Lisbon, when so many of his peers look for better opportunities in other countries, he simply replies, “I like being in Portugal”. “I feel a certain need to contribute, I have a sense of mission about it.”

July 2015