News: Inside Stories

Patrick Fernandes, a CMU Portugal Dual Degree Ph.D. student in Language Technologies (Técnico I CMU), is part of the Portuguese team that developed the European Large Language Model (EuroLLM), launched in September 2024. The initiative was being led in Portugal by their Ph.D. supervisor André Martins, a CMU Portugal Alumni, Researcher at Instituto de Telecomunicações from Instituto Superior Técnico and co-founder and VP of AI Research at Unbabel. 

The “EuroLLM: Open Source European Large Language Model” is the most advanced language model of its size developed in Europe to date. Built using the cutting-edge EuroHPC infrastructure, the EuroLLM is a fully open LLM capable of understanding and generating text across all the 24 official European Union (EU) languages, as well as 11 commercially and strategically important international languages.

Let’s take a look at what it means for a 28-year-old Ph.D. student to be involved in this major initiative.

 How did the opportunity to work on the development of EuroLLM come about?

I had just returned to Portugal after spending two years at CMU. André told me they were starting a new project for a European LLM and, since I had recently gained experience training and working with large models and distributed systems (from time doing research at Google) he asked me if I could join the project and help out.

• Could you tell us more about your specific role and contributions to this project?

I developed the main codebase and pipelines for the pre-training of EuroLLM, the phase where the model sees massive amounts of data across 30+ languages and learns its core capabilities.

It was actually developed for another LLM that we trained, CroissantLLM, where I led the training of the model. We then reused and adapted the codebase for EuroLLM, but Pedro (the lead author of EuroLLM) took charge of the training runs, and I mostly helped out discussing design decisions.

• How does your participation in EuroLLM align with your Ph.D. research?

My PhD is focused on understanding and aligning models for machine translation. Given the multilingual nature of EuroLLM (and its already exceptional capabilities at machine translation) it will probably serve as a base model for further research I will do in these last few months of my PhD. In particular, I’m currently interested in adding other modalities to EuroLLM (such as vision) and how these other modalities might help models learn new language and translation capabilities.

• How do you see the EuroLLM initiative impacting the accessibility and inclusivity of language technologies in Europe and beyond?

I think EuroLLM is a massively important step for the European AI landscape, in the currently English-&-American centered AI world.

Europe has mostly been a “follower” in recent developments with LLMs, and models like EuroLLM show that Europe can lead and innovate in this field.

It also gives us sovereignty, allowing us to prioritize not only English but all languages in Europe, and increase the inclusivity of language technologies in Europe.

• Looking ahead, how do you envision the future of this field?

I think we are finally getting over the idea that scale is all you need, so I think we will see more and more models that are smaller, but trained on very high-quality data.

But in all honesty, I think the future of this field will not be about new & better models (and whether they will replace us) but about how we can best apply current and future models to new problems and ease the interaction between humans and machines, as there is still *so much* potencial there.

• Has your experience with the CMU Portugal program met your initial expectations?

Overall, my experience in CMU Portugal has been great. I didn’t really know what to expect, but I was lucky to get a pair of advisors that work well together and provided me with a lot of support.

The experience, connections and friends I made in Pittsburgh were also invaluable to my path as a researcher (and I overall loved my time there), but being able to do world-class research while in Portugal (close to family and friends) is something I’m very grateful for.

• You are set to graduate in 2025, and the future looks promising. What are your plans after graduation?

Honestly, I wish I knew 😅. I’m definitely taking a big, multi-month break after my Ph.D. to recharge and think about what I want out of my career (and life in general).

But after that, I might pivot to working on the more applied side of AI, on applications close to my heart (such as AI for drug discovery, or better mental health treatments).

But who knows, let’s see what the future holds 🙂 

 

Hirokazu Shirado, Assistant Professor at Carnegie Mellon’s Human-Computer Interaction Institute (HCII), recently conducted a visiting research period in Lisbon. With financial and institutional support from the Carnegie Mellon Portugal Program, Shirado visited the Group of AI for People and Society (GAIPS) from INESC ID led by Ana Paiva, a researcher at INESC ID and Professor at Instituto Superior Técnico (Técnico) – Taguspark in Oeiras. Shirado worked with the group for under two months, remarking that the visit had been a fantastic opportunity to expand his research agenda and academic networks through communication with researchers at Portuguese institutions, including Ana Paiva and her lab members. 

Shirado reflected on his activities, noting that they could be summarized into three distinct areas:  (i) research development, (ii) research talks, and (iii) cultural experience. 

Throughout the visit, weekly meetings took place with Paiva and later Joana Campos and Filipa Correia, GAIPS members, to develop a collaborative research project. All parties shared an interest in the potential of machines to facilitate human collaboration and decision-making in groups. While machine intelligence has been developed for individual use, it may not always be effective in group settings due to social dynamics and network mechanisms. Shirado recently published a paper examining this point in resource-sharing (Shirado, Hou, & Jung. Scientific Reports 2023). 

Shirado also had a chance to exchange research ideas with Fernando Santos, a former GAIPS member and now an assistant professor at the University of Amsterdam, in the Netherlands. Santos and Shirado had communicated with shared interests in social networks and link recommendations before his visit and were able to further discuss it onsite, namely through their collaboration based on Shirado’s recent paper published by PNAS (Shirado, Kasahara, & Christakis. PNAS 2023). Although long-time acquaintances, this was an opportunity to deepen their ideas and explore a collaboration that now extends past the visit. 

The CMU researcher gave three research talks around Lisbon during the visit: the first talk was for GAIPS at IST-Taguspark, the second was for Zachary Mainen and his lab members at Champalimaud Centre for the Unknown, and the third was a CMU Portugal Talk at Interactive Technologies Institute (ITI). 

During the talk at the Champalimaud Centre for the Unknown, Shirado talked about his ongoing project involving neuropsychological assessments in social network experiments to Zachary Mainen, who leads a large research group about neuroscience at the medical institution. Through his half-day visit to Mainen’s lab, including the talk, they had fruitful discussions with his lab members to consider the project’s direction. Shirado also had the privilege of introducing his research to CMU Portugal and Técnico I ITI  members, such as CMU National Co-director Nuno Nunes, CMU Portugal Executive Director Silvia Castro, and Instituto Superior Técnico Full Professor, Pedro Lima (Figure 1). 

Finally, the two-month stay in Portugal allowed Shirado to learn more about Portuguese culture and triangulate cultural similarities and differences among the United States, Japan (his original country), and Portugal.

“I was amazed to see how people have similar and different social norms in cooperative contexts throughout daily life. This experience-based learning has enriched my research consideration as well as my  personal life.”, shares Shirado.

Following his visit, Shirado shared his opinion about the visit under the Program and left some considerations for future visitors.

What advice would you give to fellow CMU faculty who are considering a visiting period in Portugal?

I would suggest they enjoy different working and living experiences in Portugal. Although we can communicate with collaborators online, visiting them and talking in person could give us a different level of opportunities to foster our academic network. Fortunately, I got them through the CMU Portugal program.

How would you describe the overall outcome of this visiting period? What were the benefits of traveling to Portugal? Have the collaborations you made while visiting continued since your return to CMU? If so, please provide a brief update.

Ana Paiva and I have been working together to realize the research idea we developed during my visit. After I returned to CMU, Pavia formed a research team for the project involving Joana Campos, Filipa Correia, and their students. We are scheduled to discuss the details of the research implementation at the end of this month (February 2024). Separately, I also have monthly meetings with Fernando Santos to make concrete progress in realizing the research ideas we discussed during my stay. Actually, Santos won an ERC Starting Grant, including me as a collaborator.

Do you have plans to collaborate further with these or other Portuguese colleagues in the future?

I will continue the collaboration described above and look for further opportunities to work with other Portuguese colleagues.

More Info about the visit:

Visiting Faculty: Hirokazu Shirado, Assistant Professor, Human-Computer Interaction Institute, School of Computer Science, Carnegie Mellon University
Visiting host: Professor Ana Paiva
Visiting place: Department of Computer Science and Engineering of IST, University of Lisbon Visiting period
Period: from October 1st to November 24th, 2023 

 

Cláudio Gomes is a third-year CMU Portugal Dual Degree Ph.D. student at Faculdade de Ciências da Universidade do Porto (FCUP) and the Electrical and Computer Engineering Department from Carnegie Mellon’s College of Engineering. Currently in Pittsburgh, Cláudio shared his experience at CMU thus far including the most significant differences between Portugal and Pittsburgh and advice for new students coming to CMU.

What attracted you to the CMU Portugal Dual Degree Ph.D. Program?

The idea that I could get to know a very different country while still having a connection with my own country was very compelling. I am very particularly in favor of broadening your mind and leaving your comfort zone as much as you reasonably can. The CMU Portugal Dual Degree Ph.D. Program aligns perfectly with my vision and is a very enriching experience, both academically and personally.

What is the biggest difference between studying in Pittsburgh versus Portugal?

Definitely the campus experience. At CMU, most departments share a single campus, so it’s very easy to get to meet new people from different fields. In Portugal, universities are usually more dispersed geographically, with several campi. The cultures are also different between both sides, which makes for some (exciting) learning experiences.

What is your favorite thing about CMU so far?

Once again, I will have to say it’s the campus experience. The campus is vibrant, has a lot of life, and is full of student organized clubs, such as the CMU RPG Association. It makes me very happy to get to make friends that are not from Computer Science or Electrical and Computer Engineering.

Do you have a go-to spot on campus?

Two scoops of Millie’s chocolate ice cream at Tepper Business School is something that never fails me. I always get some every week!

What advice would you give to someone coming to CMU?

I suggest newcomers to prepare in advance for courses, to be consistent, and to enjoy their weekends as much as possible! Consistency and wellbeing are very important. Some particular things might be different, especially for Portuguese students, such as curved grades and the lack of retake exams, which can be very stressful. In practice, these things don’t matter that much, so try to not let the stress go out of control.

Is there anything else you’d like to share?

Another important piece of advice I want to give is to join student organizations! Chances are that there is a club that you will love and enjoy a lot! Search for clubs and participate in their activities if you are interested. They will be happy to welcome you!

I joined the CMU RPG Association and also am part of the first members of the CMU Photography Club, and the experience has been amazing and very enriching!

Maria Eduarda Andrada is a CMU Portugal Affiliated Ph.D. student in Robotics from Brazil. She is hosted in Portugal at Faculdade de Ciências da Universidade de Coimbra (FCTUC) and will spend her research year (spring and fall of 2023) at the Robotics Institute supervised by George Kantor. Maria Eduarda has also been part of CMU Portugal Large Scale project Safeforest which ended in 2023 and was led by her CMU host. Currently in Pittsburgh, she shared her experience  thus far and what she is looking for the rest of her stay at CMU.

Tell me a little about yourself. 

I am Brazilian and have been very nomadic in terms of lifestyle and choices. In the last 10 years I have changed 5 countries. I have always been interested in robotics, cinema and soccer. All in all, it has been an interesting life path so far and I hope it will continue for many years to come.

What attracted you to the CMU Portugal Affiliated Ph.D. Program? 

The idea of working at one of the best robotics institutes in the world with a growing and diverse environment in the Kantor lab.

What is the biggest difference between studying in Pittsburgh versus Portugal? 

I think one of the interesting aspects is the size of the lab, the social activities, and the environment. There are many more students in the lab where I work than in Coimbra, so I have more opportunities to brainstorm and network compared to University of Coimbra. The other thing is the activities on campus that are not related to my field. There are many different free activities on campus that allow me to expand my skills and interests that I did not have before. For example, Tartan Salsa, which has helped me meet new people in the city and learn a whole new skill

What is your favorite thing about CMU so far? 

I would say that discovering new activities outside my comfort zone is one of my favorite things about CMU.

Do you have a go-to spot on campus? 

At the moment, it would mostly be Newell-Simon hall.

What advice would you give to someone coming to CMU? 

Enjoy new experiences and explore as much as you can. Also, try to attend seminars and lectures, as they are usually given by incredible speakers that we would not normally have access to.

 

Manuel Reis Carneiro is a CMU Portugal Dual Degree Ph.D. student in Electrical and Computer Engineering at Universidade de Coimbra and CMU’s ECE Department. He is supervised in Portugal by Mahmoud Tavakoli and at CMU by Carmel Majidi. Manuel has an M.Sc. degree in electrical and computer engineering with a specialization in automation from Universidade de  Coimbra. His research interests include stretchable and printed electronics, bioelectronic systems, and human-machine interfaces, with a special focus on medical applications.

In addition to his academic life, music has been one of his primary hobbies, and he has learned to play various instruments, including the drums, guitar, accordion, among others.

Let’s know more about his path. 

In 2019 we published this article on our website about your research under the CMU Portugal ERI Stretchtonics. So, your connection to the Program goes way back, before your Ph.D.
Can you let us know more?

Back in 2018, I was looking for a research group in which I could develop my MS.C. Thesis, and I came across the Soft and Printed Materials Lab and their work on soft and stretchable electronics. At the time, I met with Professor Mahmoud Tavakoli and pitched to him my idea to extend the soft electronics applications that had been proposed by his lab to the field of wearable electroencephalography (EEG) which interests me a lot. By the end of my Ms.C. Thesis in 2019, which was integrated into the Stretchtronics Project, we had developed a flexible, reusable, and low-cost wearable based on printed electronic textiles for assessment of brain function and brain-computer interfaces. This wearable EEG band was distinguished  in the Fraunhofer Portugal Challenge 2019 and received an honorable mention in the IBM Scientific Prize 2019. At the same time, Professor Tavakoli invited me for a research position in the Soft and Printed Microelectronics (SPM) Lab to keep developing this technology and to collaborate with other CMU Portugal projects. 

Why did you decide to pursue your Ph.D. Degree under the CMU Portugal Program?

I first read about the dual degree opportunity after my MSC. and I discussed it with Professor Mahmoud. By the time I decided to apply to the dual degree Ph.D., I had already started my Ph.D. in the University of Coimbra, and thought that the CMUPortugal program would provide me with  valuable international experience and a distinctive chance for me to expand my horizons and establish a strong research network in both Portugal and the United States.

You collaborate closely with your Supervisor at ISR Coimbra, Mahmoud Tavakoli. Which role did/does he play in your research path?

Mahmoud has played a pivotal role in shaping my research path. His guidance, expertise, and unwavering support have been instrumental in my academic and professional development so far. Mahmoud’s role goes beyond being just an advisor. He has been a mentor, providing valuable insights, challenging me to think critically, and pushing me to explore new avenues in my research. One of the most valuable aspects of working with Mahmoud is his ability to foster an environment of intellectual freedom and encourage me to think big. He never limits my ambition, constantly pushing me to aim higher and achieve more. This mindset has been transformative in my research journey, allowing me to explore novel ideas, navigate complex challenges, and giving me as well the opportunity to fail and try again.

Currently, at CMU, you are working in Carmel Majidi’s Soft Machines Lab. How is it going?

Working at Carmel Majidi’s Lab at CMU has been an incredible experience. I have the privilege of collaborating with exceptional colleagues whose expertise spans various fields. The lab environment provides numerous opportunities to collaborate and learn from others. Carmel sets high standards and challenges us to think outside the box, stimulating everyone to push the boundaries of what we can achieve. His mentorship has been invaluable, providing guidance, support, and valuable insights into all aspects of my research.

Are there significant differences between the two research labs you work in, the “Soft and Printed Microelectronics” at ISR Coimbra and the Soft Machines Lab at CMU?

Even though both labs (Carmel Majidi’s Lab in CMU and Mahmoud Tavakoli’s Lab in Portugal) do research related to soft engineering, Carmel has been focusing more on the field of soft robotics, while Mahmoud has pivoted more towards soft electronics.

“Being exposed to such a huge variety of projects in adjacent fields and getting feedback from people across various disciplines, from chemistry to robotics or mechanical engineering, adds greatly to the richness of my experience.”

On the other hand, Carmel and Mahmoud have been actively collaborating on various projects in the last few years. As a result, there are shared aspects between both labs, for instance, in terms of printing techniques, pieces of equipment in the lab and specific materials used in some applications. This “overlap” between the SML and the SPM has been advantageous while transitioning from Portugal to CMU as it facilitated a seamless and efficient transition, minimizing the downtime in my work as I was already familiar with most pieces of equipment and techniques used in both labs.

What are the main advantages of developing your research between two countries, Portugal and CMU?

By developing my research between CMU and the University of Coimbra, I’m able to tap into the vast academic resources, specialized knowledge, and collaborative prospects offered by both institutions. One of the main advantages is offering Ph.D. students a broader perspective and enhancing their research capabilities by being exposed to different approaches, methodologies, and academic cultures in both the US and Portugal.

“This program also facilitates networking with leading researchers, professors, and industry professionals in both countries, allowing us to establish valuable connections and build a strong professional network that can contribute to future career prospects.”

Please give us a brief explanation of what you are working on under your Ph.D..

My Ph.D. research focuses on developing a miniaturized and fully integrated wearable monitoring system based on printed stretchable electronics for a comprehensive assessment of cardiovascular health. Considering that cardiovascular diseases are the leading cause of death globally, I aim to develop wireless electronic stickers capable of continuously and non-invasively measuring blood pressure and electrocardiography in real-time so that users can be monitored with clinical-grade accuracy while maintaining their daily routines outside medical facilities.

Additionally, the study involves the development of environmentally friendly and recyclable soft materials with electronic functionality to address pressing issues related to electronic waste and the sustainability of electronic devices. The beauty of new perspectives lies in their ability to unveil alternative paths.

“The Dual Degree has provided me with the opportunity to look at my research and my personal development through the lenses of two distinct cultures across two countries on opposite sides of the pond. It continues to be an amazing journey filled with growth, learning, and remarkable experiences.”

Do you already have plans for your future after the Ph.D.? Do you plan to return to Portugal?

One of my long-term goals is to secure an academic position in Europe eventually, and returning to Portugal is definitely an option I am considering. As someone with a strong connection to my home country, I believe that contributing to education and science in Portugal would be a meaningful way to give back and make a difference, with the advantage of being closer to my family and friends. However, before settling into a permanent academic position, I’m motivated to expand my research, scientific collaborations, and professional network through other international experiences. Exploring postdoctoral positions abroad is certainly part of the plans after the Ph.D.

A piece of advice for potential Dual Degree Ph.D. candidates. 

The best advice for aspiring dual degree Ph.D. candidates, and indeed for anyone, is to seize the opportunity. Embrace the challenge, and don’t hold back. This is your opportunity to push the boundaries of what you believe is possible and to leave a lasting impact in your field.
On a more practical note, make sure to leverage the support and guidance of your (potential) advisors both in Portugal and CMU, as well as the dedicated CMU Portugal staff. They are invaluable resources who will assist you in navigating any bureaucratic processes and address any challenges that may arise along the way, both during the application process and the Ph.D. itself. Their expertise and assistance will ensure a smoother journey throughout your academic pursuit, allowing you to focus on your research and personal growth and maximizing your potential for success.

Neeta Khanuja is a second-year CMU Portugal Dual Degree Ph.D. student in Human-Computer Interaction (HCI) at Instituto Superior Técnico (Técnico) and at the Human-Computer Interaction institute from Carnegie Mellon School of Computer Science.  Currently in Pittsburgh, Neeta shared her experience at CMU thus far including the most significant differences between Portugal and Pittsburgh and what she is looking forward to the most while at CMU.

What drew you to the CMU Portugal Dual Degree Ph.D. Program?

To begin with, I learned about the Program because I was already working for almost a year with Valentina Nisi on the MEMEX project, which was an EU project. When this opportunity came up and the applications opened, Valentina suggested I apply. After looking into it, I saw an opportunity to get exposure and have my research network in both places. The whole idea of being able to take advantage of two universities was something that was interesting to me. Also, in my case, I had done a Dual Degree Masters before between Germany and New York, so I knew how the dynamics would be, and I thought I would be comfortable with it. 

What is the most significant difference between studying and researching in Pittsburgh versus Portugal so far?

One of the differences for me was the language barrier. If I had to conduct qualitative research in Portugal, I would have a significant language barrier. When I was in Lisbon, it was my first year, so I think it’s more about finding a research direction and trying to understand what interests you and a lot of literature study. I think that would have been common in both places, the kind of process that you go through. 

It is also very interesting to work in Portugal because there is a very different approach to research in the way that there is a lot of philosophical inclination in HCI, so I think that is a very fresh approach. Versus, I find that it is a very applied approach here at CMU. So, I think it is also very interesting to see how these different places kind of have different perspectives. 

What has helped you adjust from living in Portugal to Pittsburgh the most?

When I came to Pittsburgh before the registration, everything was almost done. I think the Orientation week was helpful to understand a lot of things about a Ph. D. in general. Queenie Kravitz and Megan Berty provided a good support system at CMU because I had many questions about how to plan my program. We have two years at CMU, and three years in Portugal, so I was also concerned about my OPT, but I think everyone is super friendly and super open to brainstorming different ways of how things can be done. In terms of my academic experience, it was comforting to know that there will be people who will figure out a way of doing it. Personally, I find people in Pittsburgh were amicable, so it was not that difficult for me to settle in.

What are you looking forward to experiencing while at CMU?

I am really interested in contemporary dance. There is a drama and dance school here, so I’m actually looking forward to watching some of the students’ work. And the museums such as contemporary art museums. It is something I like, and I’m waiting to get some free time to explore it a bit more. 

Do you have a go-to spot to study on campus?

The Rohr Cafe in Gates Hillman Center. I have my office on the second floor. When I want to work in a café space, I think the Rohr Café is the one I prefer just to get a coffee and work.

 

Alex Gaudio is a 5th-year CMU Portugal Ph.D. student in Electrical and Computer Engineering at FEUP and CMU. Before starting his Ph.D., Alex was a Data Scientist and an Engineer in New York City for 7 years, where he co-founded the non-profit NYC Makerspace and also worked with various startups. Alex completed his Bachelor’s degree in Music at Bard College.

While in the CMU Portugal Program, he has contributed to two different projects. During his first Ph.D. years, he was involved at SCREEN DR focused on Machine Learning applied to Diabetic Retinopathy Detection. He is now engaged in CMU Portugal’s TAMI project to develop the next-generation of explainable machine intelligence in medical image analysis.

“I always wonder about the future and do things to explore the possibilities. Make some space to be reasonably safe and healthy. Challenge yourself to think bigger. A lot bigger. Consider that the pathways everyone shies away from hide opportunities you are overlooking.”

You have a curious background, you started with a Bachelor’s in music…can you share more about this part of your education?

Learning how to learn faster:  Bard College is truly an excellent liberal arts school.  I came to the school feeling intellectually much less developed than my peers.  Everyone seemed so smart.  Everyone had something to say.  People were well-spoken.  It was intimidating. Yet, when everyone is smarter and more knowledgeable than you, the world is absolutely, undeniably amazing!  The safe, supportive and stimulating atmosphere helped me learn how to thrive in life. As I grew at Bard, my underlying passion to “learn how to learn” morphed into a mantra to “learn how to learn faster.” This mantra I had, it turned out, is not just an intellectual endeavor but a physical one too. For instance, I would frequently practice my saxophone for six hours straight, skipping lunch or dinner and ignoring all bodily needs. At one point, I had such bad tendonitis in both arms that I couldn’t use a toothbrush. In my senior year concert, I almost passed out after playing a last triumphant note to a cheering crowd. Working hard demands careful attention to the needs of the body and mind. How could I align my body and mind with the will and discipline necessary to seriously pursue jazz music. That question dominated the next several years of my life as I transitioned beyond the pursuit of music into the pursuit of my career.

What took you to make such a significant change to Software Engineering?

The start of my career:  In my Junior year at Bard, I started thinking about life after college.  Professional musicians I admire often spend their first ten years working wedding gigs, in bar scenes and taking a side job to pay bills.  I didn’t aspire to any of those.  I also decided that the practical realities of the jazz musician world would limit my options and hinder my ability to grow in the ways I want to.  My “backup” plan was to become a doctor like my dad, uncle, and grandfather.  I had taken pre-med classes.  I greatly enjoyed working and volunteering as an Emergency Medical Technician on ambulances and in a hospital.  The hospital experiences were truly incredible, and multiple physicians asked if they could write a recommendation letter for my medical school application.  But I never applied to medical school.  At the time, I felt that a physician’s pathway would be restricting myself to operating inside of a large and predefined tree, with limited ability to deviate outside it.  After conversations with multiple clinicians and my family, I found myself with a backup plan I didn’t want to pursue.

My dad’s advice tends to come sparingly and in moments of need.  “I think you don’t really want to be a doctor.  Explore the world!”  he said.  I did next what anyone might do.  I google searched the industry that was most in demand.  Software engineering appeared at the top of the list, and I figured it was a good fit.  After all, I had used a Linux computer since Freshman year.

At that time, it was summer between Junior and Senior years.  I gathered the courage to ask the CTO of Bard College, Bill Terry, to hire me to work for the school and also to teach me for credit.  I greatly benefitted from this opportunity and used it to learn to hack the school’s internet, among other things.  At the same time, I also took an intro to programming class and discovered a wonderful book by Marvin Minsky called Society of Mind.   I was hooked.

From the perspective of a liberal arts college student, the change from music to software engineering was not a big change.  If history was at the top of the list in my google search, I would have taken to that pursuit in a heartbeat.

“Some people focus on skills and knowledge as an end goal.  I don’t. I point my finger in a direction and then acquire the skills I need as a consequence of choosing to go in that direction. Sometimes I go backwards, and sometimes the chosen direction is completely arbitrary and uninformed.  I point and go at it with all my will. “

Complex concepts like vision and passion come later as a result of just doing things, and skills are more like pleasant side-effects of problem-solving in some chosen direction.  Among the most valuable skills is your finger-pointing radar; try to figure out how to structure the perceived world into problems and goals that can be approached. Then, point the finger and go!  Of course, one important prerequisite is safety.

You worked in a set of NYC startups and even co-founded a nonprofit organization. Can you share more about this period?

“Comfortably uncomfortable” is a mantra I adopted from colleagues and professional mentors in the seven years between undergrad and grad school.  The phrase defines this anxious sensation that one must  always be pushing at a reasonable rate.  If you are too comfortable, the competition will swallow you.   If you are too uncomfortable, then you burn out or get injured.  I happen to be addicted to “comfortably uncomfortable.”  Startups are a thrilling pathway.  “No’s” and fighting like hell for beliefs are the lingua franca.   One day you are taking over the world, the next everything is burning and you’re crying in a hole, the third, you feel trapped and bored out of your mind.  Startups are all about a promise of success that is beautifully hard to secure and hard to define.

Ride the wave:  I want to “make the world a better place,” right? I also want to “learn how to learn faster.”  These guiding principles involve strategic decisions one has to make.  My first job after college was a software engineer for a textbook publishing company.  “Wow!” was my favorite word, and I meant it. I became friends with an intern who was at the time also a PhD student in computer science and mathematics, Professor Neil Lutz. Thanks to his inspiration, I discovered that I wanted to pursue the mathematical and scientific side of engineering.  Around 2012, data science was a brand new term. I therefore aligned myself with a goal to become a data scientist. My second and third jobs were in advertising (a brief moment of hell I am not proud of) and marketing. It’s hard to convince people that a music major with no formal training could do something a PhD might be better suited to do. By skin and teeth, it worked.  I became a data scientist and enjoyed many successes. Perhaps I sold my soul a little to ride a mainstream wave. Within a few years, I reached a local “top” of the wave, and was starting to realize my direction was off. But then what?

Ride the wave, really?  People, especially young people, tend to push for the same idea of “best.” The “wave” is a collective movement towards this elusive “top” of the world.  At some point we all ride that wave because it’s popular, straightforward and offers an easy objective.   Indeed, if you don’t know where to point your finger, then the direction of the closest “best” thing seems like a pretty good one.  Yet, pursuit of the “best” is unhealthy; it implicitly restricts our open mindedness and ability to grow.   When everyone seeks out the same “best,” we typically end up hitting average or we get better than average by working harder than everyone else and killing ourselves and our families in the process.   The wave is a useful and problematic tool.  Perhaps we sometimes need to ride the wave to learn how the world works and where we come from.  But perhaps the tool isn’t necessary.  Perhaps I over-used it.  In retrospect, riding the wave was a hard way to go, but I did rely on it when I couldn’t figure out how to get people to believe in me.

Getting people to believe in me:  Success is made by connecting with people and pointing your finger in a way that others can understand and appreciate.  When the messaging is right, people appreciate what you want and where you’re going.  When they get it, they want to help you get there, and often, they want to join you!   When the wave you ride is lacking, stop and take a look around.  The view is quite spectacular if you take a moment to appreciate it.  When you can take a moment to look around with an open and exploratory mindset, a curiously dark and scary valley invariably shows up.   That’s when you know it’s time to get off the wave and light up the valley.  The most amazing and terrifying experiences in my life were the direct result of acting on my courage to deviate from the mainstream.

At one point you decided to take a break and rethink your next steps. Why and how did this end with a CMU Portugal Dual Degree Ph.D.?

“The world is your oyster, now play!”: About 2-3 years after undergraduate school, I had a well developed walk-in closet and living room. The closet sported two and a half homemade 3D printers, a wall of electronics, and a project graveyard of sorts. The living room table was dedicated to my heavily used drill press, oscilloscope and soldering station. My computer screen mixed between papers about embedding spaces, hardware component data sheets, and random plots. At that time, I was starting to realize and accept that, in my day job, I didn’t actually believe in making prediction tools enabling marketing companies to sell fancy shoes and bags. I was especially disconcerted when a two month deployment of our latest tools coincided precisely with my sister’s sudden obsession with $2,000 hand bags and $300 shoes on a 20k a year salary. She had been targeted by our technology.

 “I didn’t know what to do next, but I knew what I didn’t want to do. I saw no curiously dark valleys in sight at the top of this wave, but it was time to hop off anyways. When I told the CEO of our company that I was going to quit, he told me it was professional suicide. Indeed, it was, and, despite the pain it caused me, it was the best thing I ever did! “

My next job was a complete failure. I burned a connection that my boss and “professional dad” had so kindly given me, and I simultaneously experienced a head injury that, thankfully, helped me lose that job faster. The next period of my life, willed unemployment for exactly six months, was the most transformative period of my life. I had built up enough income to do it.  Where would I point next? I interviewed at Google; on my way to the interview, I told my dad I didn’t want the job. Naturally, I was rejected. I played in my 3-d printer closet, and I read about ideas like Word2Vec. My wife and I decided to have a child, and the physician was like, “Dude, get a job!”

Eventually, I convinced Uber to fly me to SF, under the premise of an interview so I could talk with my “professional dad” and former boss. I didn’t tell him exactly why I was there, but he knew.

The doors of the elevator were closing as we said goodbye. Just then, a head poked between the doors just in time to say, “Alex, go get a PhD!” I fell back into the elevator and my whole body knew he was right. Thank you again, Jeremy Stanley.  I had spent much of my career until this moment carefully finding good reasons not to do a Ph.D.

In fact, around the time I discovered an interest in data science, I also developed a strong desire to pursue a Ph.D. with Tod Machover at MIT to mix music and math.  But I lost interest when a Bard professor I had worked with discouraged me from applying.  Now, I suddenly found myself pointing a finger at Ph.D.  On the turn of a time, I secured a great part-time job through a friend and mentor, I began research at Columbia with the coolest professor in 3-d printing on the planet, Professor Hod Lipson, and thanks to Hod’s initial help, I successfully co-founded a non-profit with a friend, Mohamed Haroun. The next year, I decided I would indeed apply for Ph.D. programs. That six month period of willed unemployment was completely transformative, and it set me on the most rewarding trajectory of my career. The non-profit gave me a beautiful space to build community.  We gave tools of advanced technology and education to people for free, and then showed them how to play with ideas. There is perhaps nothing more invigorating than to see people who don’t believe in themselves or their abilities suddenly reach their hand into the bag and pull something out of it.  I am very proud of this period of my life. During this time, I revised my personal mantras: “Help others learn how to learn faster.” “Stop riding the wave.” I finally figured out how to play in the world.  And play I did.

When I received an acceptance letter from CMU, I was pleasantly uncomfortable.  “Oh no!  Now what am I going to do now?” CMU presented an amazing one-time opportunity that I had taken time to consider and pursue, but the non-profit was taking off spectacularly. I did not see space for data science research in the non-profit, and my dream to create a research lab remained unfulfilled.

After soul searching, I decided to stick with my original plan and pursue the Ph.D. at CMU. It helped when my CMU advisor, Professor Asim Smailagic, introduced me to the CMU Portugal Program.  It seemed like a great way to give my family the experience of living in Portugal. I also discovered that one of my recommendation letter writers, Divyanshu Vats, was a CMU student, and his advisor, Professor José M.F. Moura, founded the CMU Portugal Program. Things therefore seemed in alignment.  The non-profit survived without me. I became excited to explore medical image analysis with Professor Asim Smailagic.

You participated in two big CMU Portugal Projects, SCREEN DR and now TAMI , can you share more about these experiences?

SCREEN-DR studied diabetic retinopathy from the perspective of automated image analysis and machine learning.  The retina is, in my opinion, the most beautiful organ of the body.  Photographs of it hide so much detail, and are telling of our health.  My uncle and grandfather are ophthalmologists, and the project therefore aligned very nicely with me.  I was ready to pursue a Ph.D. thesis to analyze fundus images, but I paused that direction when my advisors’ funding switched from SCREEN-DR to TAMI.

TAMI is a project focused on developing the next generation of explainable artificial intelligence for chest x-ray analysis, glaucoma detection, and cervical cancer.   Within the scope of TAMI, my Ph.D. thesis pursues explainable machine learning for deep learning in medical image analysis.  I focus on developing tools of compression and fixed, non-adaptive methods to expose and address bias that arises in the analysis of medical images.

“The research experience is fantastic.  I love it. My advisors are wonderful and always there for me.  My CMU advisor Professor Asim Smailagic has shaped me in such wonderful ways.  I am forever deeply indebted to him.  My Portuguese advisor, Professor Aurélio Campilho, is also very inspiring and supportive.”

The Ph.D. experience is not without its challenges, and I have sure experienced mine.  My difficulties have to do with the practical challenges of moving my family to Portugal at the height of COVID, and with the realization that the grad school system is not designed to support families.

In your opinion what are the main advantages of pursuing a dual degree Ph.D. program such as yours?

Should I try to convince you to apply to a dual degree Ph.D. program?  Yes, apply!  Should I convince you to join?  Maybe.

These are my own personal opinions at the current moment in time.

Pro:  If you’re wondering about the CMU Portugal Program, it can be a fantastic opportunity.  The dual affiliation doubles your network and number of people you can work with, it gives you more opportunities to meet people, and offers a cultural immersion experience you may never have otherwise.  My experience verifies how cool the opportunity is.

Con:  The dual degree program is not designed for families.  If you have or want to have a family during your Ph.D., consider that the dual degree program accentuates the financial burden of the Ph.D.  The Portuguese stipend is by far too small to realistically support children.  Cost of school for my two children this year, alone, will almost use up my FCT stipend.  CMU’s stipend is better than the Portuguese FCT stipend, but also too low.  Another aspect to consider is that older children may have trouble adjusting to the foreign lifestyle, and I wish in our first year we had found for our oldest son a school teacher who speaks English.  Your whole family has to be aligned with the plan to move to a foreign country and everyone needs to work consistently towards assimilating.  I don’t know others who started the program with children, and very few people start at CMU and go to Portugal.  It is not an easy or simple undertaking.

After graduation what are your plans?

I recently started developing large life-long ideas and plans.  I can’t even begin to tell you how excited I am about them.  I am forming a life-long (or 20 year) plan with three broad goals.

First goal:  Research Lab.  I have dreamed to set up a research and teaching lab ever since I learned about data science.  The lab will build on my Ph.D. work to focus on machine learning for the stethoscope and non-invasive medical devices, with emphasis on the heart, lung and the vascular system.  I want to establish simultaneous affiliation with a university, a medical school, and with industry partners.  Perhaps a dual-appointment professorship in the university and medical school is the best route towards this goal.

Second goal:  Real-world application.   I want to see my research used in the real world.   I intend to take promising research ideas from my lab, support them with investments and IP, and give diverse teams of people the power, control, support and network they need to run with it.  I would like to contribute to or create a research accelerator or incubator program that spins off startup companies and non-profits focused on machine learning, digital stethoscopes and non-invasive devices.  I also want to advise and consult for advise industry partners, implement research projects in hospitals and clinics around the world, and set up challenges for the medical machine learning community to accelerate and apply research.

Third goal:  Address the inequality gap.  I intend to build partnerships with nonprofits and investors to source, train and support disadvantaged people to work in and lead the organizations that I help spin out.  I would like to develop a sustainable way to source or train women leaders, immigrants, black people, military veterans, prisoners (e.g. from the Bard Prison Initiative), and people from my local community.   I believe a lot in diversity and am especially passionate about shattering some (but not all) glass ceilings.  I seek people who believe in individuals, not ideas.   Give individuals power and security, allow them to fail, and establish the framework to empower and fuel them when they come back for more.   I would like to spin off many companies, create many jobs, and improve the quality of health care around the world.

I am eagerly looking for the opinions of others on how to pursue any part of these three goals.  Please email me with advice or to seek collaboration!

Towards the first of these goals, I am planning to pursue a postdoc at Johns Hopkins University with Professor Mounya Elhilali.  Professor Elhilali is a world leader in the analysis of sound, she focuses on stethoscopes, and she is already very supportive.  Johns Hopkins is a leading medical institution in the USA.  There is quite a lot more I could share, but perhaps it’s better to leave it as is.  I recently started overwhelming my wife with plans upon plans.  I am very excited.

Apart from work, what are your main hobbies and is music/jazz still part of it?

It’s kind of funny to ask a Ph.D. student with kids if they have hobbies.  There’s basically no time for picking your nose.  But yes, I do have a hobby outside of kids and academic research.

There is a Spanish composer named Francisco Tarrega.  Rumor goes that, once upon a time when he was traveling outside of his homeland, and someone asked him after a concert if he missed his home and family.  His response was to compose Lágrima, which means “teardrop.”  That song is absolutely beautiful.  It was the first song I learned on the guitar.

I picked up classical guitar as a hobby after trying to convince my wife, a semi-professional singer, to play guitar.  I happen to love the instrument.  I also tinker on the piano.  I find peace and harmony in music.  It will never leave me, and I would be so fortunate if my children discover its depth.

Alex Gaudio at CMU Portugal Orientation Day 2020

John Mendonça was one of the twelve Ph.D. candidates selected under the first edition of the Affiliated Ph.D. Programs initiative launched in 2020. The Portuguese student received his MSc in Electrical and Computer Engineering from Instituto Superior Técnico in 2020, with a major in Systems, Decision and Control and a minor in Computer Engineering. During his master’s he worked with DefinedCrowd, developing methods to detect crowdsourcing fraud. Now enrolled in the 2^nd year of his Ph.D. in Language Technologies at Instituto Superior Técnico, he is working in the MAIA research project at INESC-ID promoted by the ICT Company Unbabel. In Portugal he is advised by Isabel Trancoso and at CMU by Alon Lavie at the Language Technologies Institute.

In my free time I juggle between being a full-time foodie, playing videogames, and staying up to speed on geopolitics.
I also try to run every day, which is great as it helps keep me in shape despite being a foodie.

 

You are part of the first CMU Portugal Affiliated Ph.D. Program cohort. Why did you apply to this initiative?

The original grant I was receiving did not cover my full studies so finding a grant that would cover the 4 years was a priority. Given that I was planning on going to CMU to work with my co-supervisor for a year, the CMU Portugal Affiliated PhD program was an obvious choice.

You are being supervised in Portugal by Professor Isabel Trancoso. How did this opportunity come along?

I first met Professor Isabel as the instructor of the Speech Processing course during my masters.  I’ve always wanted to work with Machine Learning and the Speech Processing course offered very interesting applications of machine learning. My Master Thesis soon followed and was supervised by Professor Isabel as well. The thesis went well, so in the end she offered me a position at INESC-ID as her PhD student.

What’s your thesis about?

The main goal of the thesis is to develop a framework to evaluate dialog systems. One of the issues of current evaluation metrics is that they focus on a single quality of dialog or on the “overall” quality, disregarding the fact that dialog quality is multifaceted (uses context, makes sense, interesting, etc.). To this end, this thesis explores how dialog dynamics intrinsically work, and propose metrics that evaluate dialog subqualities.

Your research work is being developed with Unbabel under the CMU Portugal Large Scale project MAIA. How did this opportunity appear?

At the time of my thesis final submission, the MAIA project was already ongoing, but with an opening for the work package pertaining to dialog evaluation, at INESC-ID (one of the partners of the project). Professor Isabel is also a collaborator of the project so the whole process was very straightforward.

One of the best aspects of working under MAIA is that I have the opportunity to work with other likeminded Ph.D .students
with research topics that align very well with my own.

 How is this research collaboration working out?

We help each other often and have been supported on multiple occasions by other participants of the MAIA project. We meet every week to share our progress and discuss any issues we have. All in all, this leads to a great collaborative environment, which obviously encourages our research.

What are the main advantages of pursuing a Ph.D. in collaboration with a company such as Unbabel?

Collaborating with an ICT company is always a bonus to any PhD as it helps ground our research to the real world. This is especially important if you are interested in working in the industry after your studies. Unbabel’s role in the MAIA project also includes the provision of multilingual dialog datasets, which is a very important part of any contemporary Machine Learning research.

This summer you will go to CMU under Alon Lavie supervision, what are your expectations?

Being at CMU for a year opens plenty of opportunities to engage in parallel research topics with other CMU students and Unbabel-Pittsburgh. We have a well-defined workplan to follow at CMU, following up on initial experiments conducted so far, but we expect to collaborate with other students on related topics as well. On a personal note, American culture has always been fascinating to me, so living there for a year will definitely be an interesting experience.

What advice would you leave for potential Affiliated Ph.D. candidates?

In my experience the application process is straight forward, and is very much aligned with the FCT grant. The main advice I can give is to prepare a very strong workplan with your supervisor in Portugal and at CMU.