Ignoring his friends’ advice, Karim Beguir left his career five years ago in London as a financial engineer and returned home to Tunisia to found an AI startup.
And he hasn’t looked back since.
At NeurIPS, the world’s top AI research conference, Beguir spoke earlier this month to 300 attendees at the Black in AI gathering. And his company, InstaDeep, presented two papers — a rare feat for an African startup.
“We started from literally nothing — just two laptops and a lot of enthusiasm,” said Beguir. “If told a few years ago that we would reach such a level of recognition in AI research, I would’ve considered it impossible.”
As deep learning and AI innovation take root worldwide, Africa’s AI ecosystem is gaining traction. The business value of AI in sub-Saharan Africa is forecasted to grow more than 30x over the next seven years to $46.6 billion.
Hundreds of startups already make their home in Africa — mainly in South Africa, Nigeria and Kenya — and raised $560 million in VC funding in 2017. That’s 14x the amount raised in 2012. As many Africa-based companies and researchers adopt AI technology, their growth is being fostered by NVIDIA GPUs and online deep learning resources.
AI Ecosystem in the Making
Many believe that Africa, compared to other continents, could experience greater benefit from tech innovations because there are many areas with less legacy technology in the market, which could slow the rate of AI adoption in other regions.
ABI Research estimates that 1,500 enterprises in the Middle East and Africa adopted AI in 2018 — a number that’s projected to rise to nearly 56,000 in 2022.
“The limited reach of institutional banks and hospitals across Africa has necessitated the wild growth of AI-powered mobile finance service offerings, mobile health knowledge and diagnostics platform,” said Alexander Tsado, product marketing manager at NVIDIA and director at the Africa Alliance for AI organization. “Africans at home and in the diaspora recognize this opportunity and are collaborating at an unprecedented rate to apply AI across domains.”
Startups play a key role in AI development across Africa. In areas where governments haven’t prioritized AI adoption, and cutting-edge hardware is hard to import, small companies can quickly deploy deep learning software that uses GPUs in the cloud. These startups can also serve as a training ground for young engineers who haven’t yet built AI tools.
African Startups Fostering GPU Adoption
African startups spanning industries are adopting GPUs — like Nigerian company Ubenwa, a member of our NVIDIA Inception virtual accelerator program, that detects birth asphyxia from a newborn’s cry using neural networks.
South Africa’s Cyberflex, also an Inception member, is using deep learning to help small and medium-size businesses acquire more site traffic and customers with a recurrent neural network offering powered by NVIDIA GPUs hosted on Google Cloud. Its director, Norman Faught, is a data scientist who gained his machine learning skills online through Coursera.
Moroccan entrepreneur Younes Moumen taught himself AI skills through his interest in video game development. Now the co-founder of the Inception startup ATLAN Space, Moumen’s company is using NVIDIA GPUs in the cloud and autonomous drones equipped with the Jetson TX2 module to track illegal fishing in the Seychelles.
Another startup, Nigeria-based financial technology company Aella Credit, uses cloud-based NVIDIA GPUs to power its AI. CTO Wale Akanbi, a self-taught AI developer, is building deep learning tools to determine individuals’ eligibility to borrow money and to predict their ability to repay loans.
Paving the Path for Future Leaders
From DEMO Africa to Deep Learning Indaba and AI Expo Africa, conferences focused on AI and emerging technology have sprung up across the continent in recent years. InstaDeep won best poster at this year’s Deep Learning Indaba, which also featured Beguir as a speaker.
InstaDeep, another Inception program member, works with large companies across Europe in the energy, manufacturing, mobility and transportation sectors to build AI solutions for optimization tasks.
It now has 60 employees, more than $7 million in funding and offices in Tunisia, Nigeria, Kenya, England and France. The company invested in an NVIDIA DGX-1 system and also uses GPUs in the cloud to power its work.
“We want to prove that Africa has a significant role to play in the AI ecosystem,” said Beguir, “And that you can find talented researchers anywhere in the world.”
Another poster winner at Deep Learning Indaba was South African graduate student Windy Mokuwe. Each winner received an NVIDIA TITAN V GPU to power their future work — Mokuwe’s first GPU.
All 500 conference attendees received credit towards online training with the NVIDIA Deep Learning Institute. The platform uses GPU-accelerated workstations in the cloud to give developers hands-on experience with deep learning, even without having GPUs onsite.
“These kinds of events are great in that they make us be aware of opportunities or skills that we can grab,” Mokuwe said. “It’s helping us to understand what we can be and use our degrees to do.”
Mokuwe grew up in a rural area in Limpopo, a South African province where more than three-quarters of residents live below the national poverty line.
She now studies at the University of Pretoria, but says computer science programs at most local universities don’t have courses focused on AI and machine learning. Her deep learning experience comes from an internship with South Africa’s Council for Scientific and Industrial Research.
InstaDeep’s Beguir, who grew up in Tunisia but spent a decade working in New York and London, sees his return to Africa as an opportunity to mentor young engineers in the importance of AI adoption.
“It’s important that we break the psychological barrier in Africa that good things happen elsewhere,” he said. “It’s essential to bring back hope by creating success stories. A few positive examples would go a long way.”
Beguir says he finds it fulfilling to see his journey resonate with the young people he mentors. And, he notes, it’s still early days.
“The ecosystem is literally being built in front of our eyes,” he said. “We have a chance to steer the AI opportunity in Africa towards a positive outcome.
* Main image for this story shows members of InstaDeep’s Nigeria team in their Lagos workspace, including co-founder Karim Beguir (third from right).