Every idea Dr Sunday Oladejo works on begins with a simple question: can everyday behaviour, from climbing a hill to drifting off to sleep, be turned into a pattern that makes sense to a computer? As a lecturer at Stellenbosch University’s School for Data Science and Computational Thinking, and the chair of the Standard Bank Lab for Recommender Systems, he builds algorithms that mimic how people move, rest and respond to the world. “If you have a curious mind, you will always want to solve problems. But how can you use AI or data science to do it?” he asks.

Walking from his flat in Stellenbosch’s La Colline to campus every day, Oladejo often looked up at the hill above Kayamandi, and wondered how much effort it would take to reach the summit. As a keen hiker, he started to turn the problem over in his mind, leading to what he calls the Hiking Optimisation Algorithm, or HOA. Many models that estimate walking effort use what’s known as Tobler’s hiking function, which predicts how the angle of the slope affects a hiker’s speed. But Tobler’s model was developed before GPS data and real terrain analysis were easy to combine. Oladejo’s version builds in variables like terrain roughness, the way real people vary their pace when the incline shifts, and how hikers adjust for energy loss on steep climbs.