Top SA data scientists make their mark

Kimberly Taylor, Amira Abbas, Dr Mark Nasila and Dr Benjamin Rosman are solving critical challenges in society and achieving global recognition.
Peter Alkema
By Peter Alkema
Johannesburg, 07 Jun 2022

Top South African data scientists are solving critical challenges in society and making their mark on the global industry. From deliveries to cyber security, advanced analytics is driving change; but it’s the leading thinkers behind these solutions that are really impressive.

Delivery tech increasingly data-reliant

If you’ve ordered a delivery from South Africa’s largest on-demand grocery delivery service, the driver’s route was optimised by a “travelling salesman” algorithm that Kimberly Taylor originally developed as a Wits engineering student.

She has since built a company and an award-winning app around this innovation which helps logistics companies scale their delivery volume. Multi-stop route optimisation is critical for perishable deliveries and through Taylor’s solution, data science is helping on-demand delivery companies in the Quick Service Restaurant and grocery space keep their promise of 30 − 60 minutes.

This is why legacy industries like supply chain and logistics are turning to data science to optimise their operations. DHL's Smart Truck system optimises routes on last mile deliveries and has improved customer service and reduced costs by 15%. UPS delivers more than 21 million packages a day and uses machine learning and multi-model forecasting to analyse 5 petabytes of data per week.

These industries are actively recruiting data scientists; globally vacancies have grown 480% since 2016 and data science careers are becoming increasingly lucrative.

Quantum leap for machine learning

Durban-born Amira Abbas is an IBM research scientist, specialising in machine learning and quantum computing. As the first African to receive Google’s PhD Fellowship award in the category of quantum computing, she is an inspiration to young people across the continent who want to pursue maths and science careers.

Quantum computers are far more powerful than traditional computers because they use phenomena in quantum physics like qbits and entanglement to perform data operations. Harnessing this vastly increased computing power for technologies like artificial intelligence and machine learning offers enormous advantage.

Without sufficient local research, technologies like AI may further perpetuate the imposition of external values on Africa.

Through her work with IBM’s quantum computing research lab in Zurich and the Quantum Research Group at the University of KwaZulu-Natal, Abbas hopes to ensure this computational leap will solve real problems for people. The idea of “quantum enhanced” algorithms could combine the intelligence of machine learning with the power of quantum computing.

Cyber-defense is the new battleground for AI

Chief analytics officer for FNB Risk Dr Mark Nasila is tasked with using data science to defend the bank against AI-powered attacks. As hackers have become more sophisticated and tech-savvy, so the bank’s cyber-defense has ramped up the use of AI and machine learning in areas such as risk management and forensic due diligence. Huge datasets and complex variables can quickly and accurately be analysed for patterns to detect would-be intruders.

More advanced forms of AI like machine learning are deployed against DNS tunnelling; they can also detect malware and even identify and defend against zero-day exploits. Much like a biological virus is killed off by the immune system, so too this intelligent cyber warfare displays evolutionary tendencies in how these models learn, infer, predict and ultimately defend.

Demand for data scientists with cyber security training that can build these models is ramping up and the AI cyber security industry is expected to grow to $38 billion by 2026.

African champion for unbiased AI

Deep Learning Indaba founder and Wits Professor Dr Benjamin Rosman is a leading African voice in the field of ethical AI. His research argues for machine learning systems to be aligned with human values that are compatible with the societies in which they are used. He points to known biases in AI technologies such as facial recognition, which has been shown to be less accurate in identifying women and people of colour than white men.

Without sufficient local research, technologies like AI may further perpetuate the imposition of external values on Africa, a continent deeply sensitive to outside influence. The Deep Learning Indaba, and its satellite events in 27 different African countries, demonstrates the widespread interest in doing this work. Rosman also hopes for more African representation at global AI conferences and points to the UNESCO draft recommendation on the ethics of AI which calls for emerging technologies to “benefit humanity as a whole”.

Data science continues to grow rapidly as firms race to take advantage of big data. Learn Python and machine learning today and make your mark on our data-driven world.

* Peter Alkema helps lead digital transformation at FNB. This article is written in his personal capacity.