Born in Greece, at four years old her parents moved the family to Madrid, in the hope of securing a better future. After a couple of years there, she relocated to Brussels where her parents took up posts at the European Commission and where she attended a French school, the Lyçée Français Jean Monnet.
With uprooting steeped in her DNA, deciding to move to London and study at Imperial aged 17 was probably less daunting for Phebe than for many.
I was really attracted to the opportunity of a very hands-on approach to learning such as the one offered by Imperial. I had developed a real passion for math and physics in Belgium and was eager to apply my knowledge to solve practical problems. In fact, I started doing research and published my first paper during my undergrad at Imperial. This experience instilled in me a love for doing research and motivated me to do a PhD.
"It helped that London is such a beautiful city, and that there’s so much diversity and culture. I loved spending time in Soho and visiting Tate Modern. It’s also a perfect location for someone who loves alternative rock as much as I do!"
Towards trustworthy decision-support tools
At Imperial's Department of Computing, Phebe dedicated her PhD to the design of robust data-driven decision-support tools based on rigorous mathematical models and reliable algorithms.
My goal was to design reliable decision-support tools that users can really trust and that are guaranteed to perform as planned and intended when deployed, even in the presence of uncertainty (so-called “known unknowns”) or unmodelled phenomena (so-called “unknown unknowns”).
At the time, people outside the sector were not yet worried about the potential unintended consequences of deploying algorithms trained on historical data in the open world. However, the past few years have seen an explosion in the use of automated decision-support tools and, more generally, artificial intelligence (AI) in our day-to-day lives, raising growing concerns about its impact.
"I am so happy to have studied such important and timely questions in my PhD. The more we see AI being deployed and used in ever more high-stakes settings (e.g., for policymaking, to decide who to hire, or the best treatment for a particular patient), the more this work becomes relevant," she said. She then added: "When you decide on a PhD topic, it’s important to pick something that aligns with your personality and that you are passionate about. I am lucky that in my PhD I found just that: being risk averse, having tools that can help me reliably make good decisions even in adverse circumstances suits me naturally."
Long lasting connections
Phebe says she loved her time in London and at Imperial. "This is where I got to meet some of my closest friends, many of whom I still see or talk to weekly or even daily and many of whom I still meet in Europe in the summers or throughout the year at conferences." In fact, this is where she met her now husband almost 20 years ago, when she was an undergraduate in Electrical and Electronic Engineering and he was doing his PhD in the same department.
Some of the smartest and nicest people I know, I met at Imperial. Many of the people I did my PhD with went on to become incredible researchers and faculty members at top universities across the world. I continue to collaborate with many of them to this day. It is fantastic to work with people you genuinely like and admire, and that you have known for so long and know you can trust and rely on.
Only recently, she had a paper accepted for publication with one of her academic siblings at the AAAI Conference on Artificial Intelligence, one of the top conferences on AI, in the special track on Safe, Robust, and Reliable AI.
Phebe says her PhD advisors at Imperial, Professor Emeritus Berç Rustem and Professor Daniel Kuhn (now at École Polytechnique Fédérale de Lausanne), had a profound positive influence on her. "They taught me so much about how to be a good researcher and a good mentor at the same time. They showed me that you can have the highest standards in research while still creating a positive atmosphere where students feel comfortable to ask questions and learn." She continues to reach out to them regularly about research and career advice.
Getting to the University of Southern California
While she absolutely loves doing methodological research, Phebe rapidly realised that she wants more: in addition to wanting to gain a fundamental understanding of problems, she wants to ensure her solutions are put to direct use to make a positive impact on society. She had this realisation during her postdoc at MIT in Boston while creating data-driven decision-support tools for transplant surgeons at Massachusetts General Hospital. She was helping them predict, from historical data, how long it would take their patients on the very long US kidney transplant waitlist before they receive an offer for a kidney. Such predictions are useful for disease management but also to help them decide whether they should accept or reject an offer for a kidney of more marginal quality.
I find the idea of using my research in maths and computing to provide reliable data-driven recommendations for decisions that have the potential to have a huge positive impact on people’s lives extremely rewarding. It perfectly blends my love for heavily quantitative and computational work with my deep desire to help improve people’s lives and livelihoods.
This realisation attracted her to join the University of Southern California (USC) in Los Angeles. "The main reason I chose to join USC is that it promotes interdisciplinary research and collaborations between faculty in different schools. In addition to having a top engineering school, it has extremely strong schools of social work, public policy, and medicine. This is how I knew it was the ideal place for me to successfully realise my research agenda."
Another reason why Phebe chose USC is its WiSE (Women in Science and Engineering) programme, an initiative aimed at increasing the representation of women in science and engineering at USC. "In academia, in particular in engineering, you know there will be challenges you have to face as female faculty. Knowing there is a support network to help you succeed is extremely important. The WiSE programme has really been instrumental to my success." In fact, Phebe held a WiSE Gabilan Assistant Professorship from the programme between 2021 and until tenure, a distinction that serves to recognise academic excellence and the advancement of the goals of WiSE.
Elevating the power of artificial intelligence
Upon arriving in LA, Phebe realised the profound homelessness crisis affecting her new city, a city that she was growing increasingly fond of. After learning about the dramatic shortage of housing resources to help support those experiencing homelessness, and about the severe inequities in homelessness, she decided to concentrate her efforts on designing AI that can support the most vulnerable, marginalised, and underserved communities with focus on equity. Problems faced by these communities are exactly the kind that motivate her the most in her research – problems that are affected by uncertainty and ambiguity, and where data may be biased or scarce.
Importantly, it rapidly became clear that existing AI tools cannot be simply applied in these new contexts (due to, for example, risk of perpetrating historical biases, lack of transparency) and that AI research would be needed to effectively address them. This motivated Phebe to help build the USC Center for AI in Society (CAIS), which she now co-directs.
CAIS is an interdisciplinary research centre between the schools of engineering and social work at USC, the first such partnership in the world. The centre currently involves 50 faculty members, 24 PhD students, and several undergraduate and master’s students whose research is aligned with the CAIS mission and vision. The centre faculty and students routinely publish work at the interface of AI and society and regularly organise events showcasing the power of AI to tackle pressing societal challenges.
Through my activities at CAIS, I am helping define and grow the field of AI for Society, by training researchers, by organising collaborative events, by elevating researchers and communities doing work in this area, and by identifying and breaking barriers in conducting this kind of work. We do need more experts in AI helping solve these important problems. I hope that through our work in the centre, we can get more researchers excited to help and more communities interested in using AI to solve their problems.
At USC, Phebe conducts research in AI to help mitigate homelessness, conserve biodiversity, and prevent suicide and substance use. Notably, she has partnered with the Los Angeles Homeless Services Authority (LAHSA), the entity charged with allocating scarce housing resources to those experiencing homelessness across LA County. Through this partnership, she is having a direct positive impact on policy practice and society by helping design algorithms to allocate these scarce housing resources more efficiently, fairly, and transparently. In fact, she and her team recently released a policy recommendation report summarising the findings of a three year interdisciplinary effort.
She has also partnered with Panthera, an organisation dedicated to saving wild cats, and is devising new methods in AI to help save the jaguar in Latin America before it becomes extinct. Her research has been covered in the popular press, including in yahoo!news, VentureBeat, KCAW, and NPR Marketplace. Recently, she was a speaker and panellist at the inaugural TED AI event, alongside some of the most prominent figures in AI. For her work, she has earned the prestigious NSF CAREER award from the National Science Foundation, one of the biggest honours for faculty in the US.
Models and algorithms for trustworthy AI
Building upon the work she did at Imperial and motivated by problems she identifies in her collaborations with policymakers and community partners, Phebe devises optimisation-based models, algorithms, and open-source software for trustworthy AI and machine learning (ML). In particular, she has led the creation of several open-source packages for learning robust, interpretable, and fair decision-making and machine learning models that organisations and other researchers can use and build upon to help guide decision-making in high-stakes settings.
In high-stakes, low-resource settings such as the ones I have focused my research on, being open about the models, algorithms, and code is critical. We not only want the outputs of our algorithms to be transparent, we want the whole process, end-to-end to be easy to improve and scrutinise.
A passion for teaching and mentoring
"I am really fortunate to work with incredible students who are really strong technically and have a deep care for social issues and a passion about using data, models, and algorithms to address them," she said.
Phebe’s students have earned extremely competitive awards – notably one of her PhD students earned the Grace Hopper Celebration Student Scholarship and three of her PhD students have won the Graduate Research Fellowship Program (GRFP) award from the National Science Foundation, one of the most competitive awards for graduate students in the US. "This feeling that you have positively influenced a young person’s life, there is nothing quite like it," she said.
Mentoring students and seeing their fantastic achievements as researchers, teachers, and mentors, is one of the most rewarding aspects of being a professor.
Promoting diversity, equity, and inclusion
Phebe leverages her research to drive outreach activities that promote diversity, equity, and inclusion, to attract more students, in particular those with non-traditional backgrounds or from underrepresented groups to engineering and computer science. For example, she used her INFORMS Diversity, Equity, and Inclusion Ambassadors Program Award to conceive and run the ExplOR event for high school students, a day-long programme aimed to bring STEM education to underserved communities. She also hosted the USC STEM Spotlight for students of the STEM Academy of Hollywood, a public school that primarily serves underrepresented groups.
As a professor, I feel that it is my responsibility to promote DEI in our programmes, to ensure that everyone knows about what engineering and computer science can do and to get young people excited to study it.
Follow Phebe's story:
Imperial's Alumni Awards recognise the outstanding achievements of our alumni community and the variety of ways they are making a real impact across the globe.
The Emerging Alumni Leader Award celebrates our rising stars, innovators, game-changers and future leaders.