UmojaHack Africa: Reflections from AI Tanzania Community
Africa’s first and biggest virtual Inter-University Machine Learning Hackathon.
UmojaHack Africa event organized by Zindi on 21st March 2020 was the largest inter-university machine learning hackathon. The AI community in Tanzania with over fifty (50) students from universities of the Nelson Mandela African Institution of Science and Technology (NM-AIST), University of Dar es Salaam and University of Dodoma participated in the UmojaHack Africa.
“Umoja” (a Swahili word) means “Unity” in English, the national language of Tanzania (Kiswahili) was highly reflected from the focus and commitment of students participating at the hackathon despite the challenge of working from home due to the COVID-19 pandemic. The initial plans for hosting the event at a big room for the full day had failed. We appreciate the support from our sponsor Alliance For AI who supported the NM-AIST university representative, Dina Machuve until the last minute making sure NM-AIST participants at remote locations take part in the hackathon. The last-minute checks by emails and calls by Alex Tsado, founder of Alliance4ai meant a lot to us organizers on the ground and motivated me, the Zindi Ambassador in Tanzania and my colleague, Dina to pull together the students. The technical support from the Zindi team in South Africa was very helpful. With my role as the Zindi Ambassador for Tanzania, I had the obligation to prepare teams at the different universities for the event well in advance. I co-organized mini hackathons at universities to develop skills and interest in machine learning and data science.
The AI Tanzania community had humble beginnings since 2016 and has had a steady growth in both numbers and interest in the field. Three students attended Data Science Africa (DSA) event at Makerere University in Kampala, Uganda in 2016, one Ph.D. candidate attended the Deep Learning Indaba (DLI) in 2017 in South Africa and two at the 2017 Black in AI (BAI) workshop in California, USA. The numbers since then have grown to ten folds in attending regional and international AI conferences. We have co-organized two IndabaX events in 2018 and 2019 in Tanzania with over 70 and 100 participants respectively. The outcome of the active growth in the AI community includes Masters and Ph.D. research projects in machine learning and summer schools in machine learning for undergraduate students.
UmojaHack Africa was a great milestone for the AI community in Tanzania that is quite small compared to the other communities in the region such as Data Science Nigeria (DSN). The winner of the 3rd position at the Umojahack #2: Xente Purchase Prediction Challenge was Michael Nkotagu, a Masters's student at the NM-AIST. He really motivated members of the community and proved that it is possible to win at the hackathon. We interviewed some of the participants after the event and they shared useful insights.
Michael said, “It was a wonderful experience participating in the hackathon. The biggest takeaway from the event is that every second matter, and the most important thing before starting to attempt any of the given challenges is making sure you have a clear understanding of what is required and before devising a clear and thorough plan to tackle the problem before you start writing any code.”
The majority of the students took part on the Umojahack #3: Hotspots Fire Prediction Challenge. “It was our first time participating in a Hackathon, the challenge was a great learning experience in terms of skills and the knowledge required to solve the problem. There were over 150 teams/individuals working on the Hotspots fire prediction challenge and our team (Bienvenido) was ranked at 29. This was a great achievement for our team and we hope to improve and possibly win future challenges. It was a great learning experience.” This was the feedback from the team of Wasiwasi, Hope, Augustine and Mollel who worked as a team.
“I had a new experience participating in the international hackathon, I learned a lot of things considering I am a beginner,” this was the feedback from Abubakar, a University of Dodoma student. The few who attempted the computer vision challenge had a tough experience on too low internet bandwidth to access the AWS and really hoped they had a GPU locally. In general, the students acknowledged the role and commitment of Zindi in AI skills development to youths in Africa. Mbonimpaye from NM-AIST said, “I actually gained a lot. My takeaway is to do a lot of exercises on machine learning challenges.”
“I have little experience in Hackathon as so far I had only done the Zindi ArushaHackthon held in our Institution before participating in UmojaHackthon. I ranked 6th position at the Hotspots Fire Prediction challenge but I lost thereafter in hyperparameter tuning”. This was the feedback from Beston.
As a next step, we are now considering the merits of starting an Alliance4ai chapter in NMAIST and UDOM to leverage the self-learn AI toolkit from Alliance4ai and its network of fellow learners across Africa to grow as leaders of change that matters.
We, the representatives of the community are motivated and proud of the morale of the students. Our last message to them is the quote by Nelson Mandela “I never lose, I either win or learn”.
Thanks to all sponsors Zindi Africa, Microsoft, African Bank, Google AI, Liquid Telecom, Alliance4ai, Instadeep, The Field Institute, and Data Science Nigeria.
This article is prepared by Dina Machuve and Davis David.
One last thing: Read more articles like this in the following links.
