By Li-Ming Pan (Communications and External Relations, Harvard Center for African Studies)
We caught up with Harvard South Africa Fellowship Program (HSAFP) alum, Vukosi Marivate, who participated in the Program for Leadership Development (PLD) at the Harvard Business School in AY20-21.
Vukosi was recently promoted to Associate Professor at the University of Pretoria and is also the ABSA Chair of Data Science. In addition, he is the Principal Investigator and Lead of Data Science for Social Impact Research Group and founding member of Masakhane, a Pan-African research project to improve how dozens of languages are represented in the branch of AI known as natural language processing. He is also a co-founder of Deep Learning Indaba, an organization which aims to strengthen African machine learning. Beyond just doing the research, Vukosi is very much involved in building communities in research that are more sustainable. His Data Science for Social Impact Research Group has been collating COVID-19 data from the South African government daily for researchers, innovators, and developers to use. Vukosi completed his BSc and MSc in Electrical Engineering at the University of the Witwatersrand and his PhD in Computer Science at Rutgers University.
Q: What was your motivation to apply for HSAFP?
A: During my PhD in Machine Learning, looking at how machine learning and data science can disrupt the industry, I thought, one of the things that's very important in being a teacher in the research in this area is also to understand business processes and how data science and machine learning is used within a company. I was able to understand more of what is going on in business as a scientist and an academic. In addition, the PLD program gave me leadership skills such as how to affect change in an organization.
Q: Your program in AY20-21 was virtual. How was that experience?
A: I was not able to have the Harvard campus experience as my HSAFP experience was online. But I also teach, so I know that we are all doing the best we can during this time. The PLD program exceeded my expectation for what was possible in online teaching. I felt like I was in the lecture hall being able to interact, and I really enjoyed the material.
Q: What was a highlight of your HSAFP program?
A: One of the components of the program that should really be highlighted is having a coach that you can interact with and work with throughout the whole period, which was really awesome. Having the opportunity to sit with someone and think through the changes you want to make in an organization was invaluable.
Q: Since you’ve completed the program, can you share what you have been working on?
After the program, I returned to the University of Pretoria to continue my academic career and I was recently promoted to Associate Professor. A group that I lead at the University of Pretoria, the Data Science for Social Impact Research Group, has been working on the only open repository of COVID data for South Africa to get researchers, innovators, and developers access to COVID data to then use it in an open manner. The government releases PDF documents with infographics and other numbers on COVID but that is not useable for researchers, so since March 2020 we have been collating COVID-19 reporting data from the Department of Health and National Institute for Communicable Diseases and updating a spreadsheet that is machine readable. Other researchers can automate their systems to do their analysis and build their models used to inform decision makers and understand the spread of COVID-19 and the impact of various interventions. Over 60 volunteers have worked on this project since March 2020 to make sure the data is made available and cleaned up. This data issue is not only a South Africa problem, but the way many people share data is not the way we can use it to find patterns. See the COVID-19 South African Dashboard Vukosi’s team has been working on here.
I also work in natural language processing for local African languages and have been thinking about this since my PhD. It’s important, because we are getting to a point where if a machine can’t understand your language then in a few years from now it’s going to be like that language never existed. Learn more about this initiative here. This initiative also won a Wikimedia Foundation Research Award in May 2021.
I also am part of data science societies that try to make sure data is used more than just in science. I am part of a Grassroots Indaba (Deep Learning Indaba), the largest grassroots AI organization in the world based on the African continent working to strengthen African machine learning. We have had meetings in Johannesburg, Nairobi, and this year will be in Tunisia. Of course, it’s important to consider how we live with the pandemic and making sure everyone is safe and still able to be innovative and connect during this time. I think it’s important that Africans shape and own machine learning and find a community where they can have support for the development of these areas in Africa rather than having to look to the West for this support.
Q: Congratulations on your recent promotion to Associate Professor at the University of Pretoria. What is your experience with academia in South Africa?
A: My path isn’t the common path into academia as I went into government research after I got my PhD instead of staying in the academic sphere. When I went back into academia, I needed to start building up networks and research collaborations across the continent and beyond that will continue as I work towards a full professorship. I continue to be a part of the national Future Professors Programme that aims to develop the qualities of academic excellence and leadership in university scholarship in order to contribute to the development of a future South African professoriate.
Q: Do you have any advice for HSAFP applicants?
A: Think about what you want out of the fellowship. I applied to the program twice, and when I didn’t get in the first time, it did give me time to pause and think a little bit more about what I wanted out of it. By the time I got accepted to the fellowship, I knew where I wanted to go. Being a data scientist requires you to be more multidisciplinary, so I work with a lot of people outside of computing or engineering. I appreciate being in flexible positions. I looked at how to bridge the gap between how organizations work and the science that I am doing, and I knew the PLD would help me gain those skills.
Thank you Vukosi for taking the time to share your achievements and work with us! We look forward to seeing your future endeavors post-HSAFP!