Hackathons part of plan to train one million data scientists

An inter-university hackathon in which 1,150 students from 126 universities based in 21 countries participated to solve three real-world machine learning challenges is part of a plan to train one million data scientists in Africa who could become the backbone of the continent’s efforts to become globally competitive in artificial intelligence (AI).

The hackathon, UmojaHack Africa 2021, is one of the biggest regional events of its kind, aimed at improving the practical skills of data science students by creating an opportunity for them to participate in AI projects.

It was organised by Zindi, a data science competition platform that already includes a community of 24,000 data scientists who aim to contribute to solving societal challenges through machine learning and AI.

Tony Mipawa, a data science student from the University of Dodoma, Tanzania, is part of this community. A year ago he was a data science novice until he participated in Zindi’s first mentorship programme in 2020.

A year later, he was placed second in the hackathon’s Sendy Delivery Rider Response Challenge. It aimed at creating a machine-learning model that would improve the experience of customers with delivery needs by linking them to the closest available drivers.

“I’m very happy with the outcome. My advice is, whenever there is an opportunity to learn, you should take it. Learning is all about passion. Whenever there is an opportunity to learn, put your whole effort into it, do it well. Try to learn from anyone you meet,” said Mipawa.

Based on Mipawa’s experience, Zindi is accomplishing what it set out to achieve.

Celina Lee, co-founder and CEO of Zindi, said the goal is to make AI accessible to everyone, including young people pursuing data science and machine learning as career paths.

“The problems Africa faces are unique. We believe that data scientists bring the perspective and lived experience needed to design the best possible solutions for Africa,” she said.

“Our long-term expectation is for Zindi to host the data science ecosystem in Africa. We want to help create one million data scientists and for Africa to become globally competitive in a diverse and inclusive global AI market,” she added.

According to Joyce Nakatumba-Nabende, the head of Uganda’s Makerere University’s, AI and data science research lab, there is growing interest in machine learning in Africa and university students could leverage different competitive platforms to grow their skills and capacities.

“The crucial value of machine learning and data science initiatives is that they are helping to solve real issues occurring on the continent, such as sexual violence and language translation,” she said.

Real-world applications

The UmojaHack Africa 2021 featured three challenges, and 8,500 submissions from universities in Algeria, Benin, Cameroon, Côte d’Ivoire, Egypt, Ethiopia, Ghana, Guinea, Kenya, Malawi, Morocco, Nigeria, Rwanda, Senegal, South Africa, Sudan, Tanzania, Tunisia, Uganda, Zambia and Zimbabwe were received.

The beginners challenge focused on a task that required students to build a machine-learning model to predict which individuals across Africa and around the world were most likely to be financially resilient. The prize was awarded to a team of students from the University of Lagos in Nigeria.

Jomo Kenyatta University of Agriculture and Technology in Kenya topped the intermediate challenge on logistics, the Sendy Delivery Rider Response Challenge, aimed at creating a machine-learning model that would improve the experience of customers with delivery needs by linking them to the closest available drivers.

A team of students from Tunisia Polytechnic School (Ecole Polytechnique Tunisie) (EPT) won the advanced challenge that required students to evaluate neutralising antibodies for the next influenza pandemic using the DeepChain platform. The platform helps biotech companies and individuals, through AI, to advance a basic understanding of protein function.

Complex systems engineering student at Carthage University (Université de Carthage) in Tunisia, Hazem Barka, who was part of the winning team, stated that data science and machine learning have the power to modernise economies in Africa and the competition opened up working and internship opportunities for students.

“The event made us realise that AI can help to save people’s lives by accelerating the antibody effectiveness evaluation,” he said.

Mohamed Amine Mairech, a third-year student at EPT specialising in machine learning and data science, added that the event had allowed students to network, connect and share knowledge and ideas within the AI community in Africa.

“The competition was a great opportunity to learn about a new domain, which is biology, and to see the application of AI in such a complex domain,” he said.

The winning models developed by participants are to be shared with relevant organisations and used in real-world applications.

Paul Kennedy, a spokesperson for Zindi, told University World News that many students across Africa pursuing data science and machine learning lacked exposure to competitive platforms necessary for skills development.

“Data science is quite a difficult field to break into if you do not have any practical experience, and there only a fairly limited number of formal education opportunities in the machine learning sectors,” he said.

“Students would need to take up maths, statistics, computer science and engineering, and find ways to grow their practical skills outside of the university [context].”

The hackathon was funded by AI and financial organisations, including InstaDeep, the Standard Bank Group, DeepMind, Microsoft, NVIDIA and Old Mutual.

Women under-represented

Despite the success of the hackathon, the number of female participants at the event remained resoundingly low at 21%.

Similarly, out of a community of 23,000 Zindi users, only 25% were women. This aligns with findings in a UNESCO report, The Race against Time for Smarter Development, published in February, about women in the digital economy. The report pointed out that women were excluded from the digital revolution.

Commenting on the gender disparities, Lee stated that, for AI to improve lives in Africa, it was critical to have more women around the table designing solutions, in particular since they were one of the groups most affected by AI and applications.

“I believe we have to explicitly combat [women’s lack of representation] by making a commitment to representation at all levels, in all domains, from educational environments to places of work – so that the culture around data science changes and is more welcoming to women,” said Lee.

Nakatumba-Nabende also highlighted the need to increase females’ enrolments into machine learning and AI fields through mentorship programmes and scholarships.

“In our masters in computer science at Makerere University, we have very low participation of women. For example, in a class of 24 students, there are only two female students.”

To support women participation, Zindi also established a mentorship programme specific to women data scientists in addition to competitions and hackathons targeted at women in machine learning.

In April 2021, Zindi hosted a “Women’s hack for safety” initiative to help combat gender-based violence against women by using machine learning to predict which women are at the highest risk of becoming victims in South Africa.

Said Lee: “Universities and governments must consider incentives that explicitly support women in building their skills, practical experience, networking and career opportunities in data science.”

AI labs and communities

Beyond hackathons, the establishment of AI labs at different universities across Africa has helped to bridge the skills gap in machine learning.

The Makerere University AI lab has been helping to advance AI by providing internships for students and offering a six-week introductory course for beginners.

Nakatumba-Nabende mentioned that different academic institutions in Africa were establishing AI labs to allow learners to interact with real data sets collected by universities and to receive training on building models.

The emergence of AI and data science communities has also been a turning point in drawing out talent, bringing together data enthusiasts and mentors to share knowledge on data usage.

The Masakhane project, for example, designed to translate African languages using neural machine translation, brought together AI professionals based at universities, students and local communities to generate data and interact on different aspects of machine learning.

Several data science initiatives, such as the DataFest Kampala and Hackathon on Economic Well-being Prediction Challenge developed by the African Institute for Mathematical Sciences, known as AIMS, have created opportunities for students to sharpen their competencies in data science.