Witsies awarded prestigious Google PhD Fellowships
- Wits University
Jess Rees and Tristan Bester, have been awarded the highly competitive Google PhD Fellowship for 2025.
The fellowship recognises exceptional PhD students from the School of Computer Science and Applied Mathematics (CSAM) at Wits whose research demonstrates the potential for groundbreaking scientific contributions and meaningful societal impact, particularly in areas where advances in computing can benefit communities around the world.
The Google PhD Fellowship programme supports outstanding graduate students pursuing innovative research in fields relevant to Google, including machine learning, artificial intelligence, and their applications. Fellows are selected through a rigorous international process and are recognised not only for academic excellence, but also for the ambition of their work to address real-world challenges and shape the future of technology.
Advancing AI for healthcare: Jess Rees
Jess Rees is supervised by Distinguished Professor Bruce Bassett from CSAM, and has been recognised for her research on the use of AI in medicine, with a particular focus on improving clinical diagnosis and strengthening healthcare systems in South Africa.
Reflecting on the award, Rees said she feels “excited and honoured,” describing the fellowship as a powerful validation of the potential impact of her research. “The possibility of influencing how AI is developed for healthcare, and of making a large, positive contribution to the South African healthcare system, is hugely motivating,” she said. “I feel even more driven to contribute to both the field and the country through my work.”
Rees highlighted the freedom the fellowship provides, allowing her to focus fully on her PhD research and to build collaborations across the CSAM, the Faculty of Health Sciences, and the Machine Intelligence and Neural Discovery (MIND) Institute.
She also expressed enthusiasm for the mentorship opportunities offered through Google and the global community of PhD fellows, adding that she is eager to “dive in and nerd out as we develop new algorithms for clinical diagnosis.”
Previously a Principal Data Scientist at Discovery Health, where she worked for nine years, Rees has transitioned to a full-time PhD in Computer Science at Wits. Her career has been rooted in natural language processing, with recent work centred on large language models and generative AI.
In 2024, she was named one of the Mail & Guardian’s “200 Young South Africans” for her contributions to Technology and Innovation.
Her doctoral research will focus on multi-agent AI systems for sequential clinical diagnosis, aiming to deliver practical, high-impact tools for healthcare.
Strengthening the foundations of AI: Tristan Bester
Tristan Bester received the Google PhD Fellowship for his work in machine learning foundations, particularly at the intersection of reinforcement learning and symbolic AI.
“I am honoured to receive the Google PhD Fellowship for Machine Learning and ML Foundations”, Bester said. “It is truly a privilege to have an institution as prestigious as Google support my research and recognise the importance of my work. This motivates me to continue pushing the boundaries of AI and machine learning.”
Bester is a PhD candidate in Computer Science at Wits, where he is supervised by Professor Benjamin Rosman, Director of the MIND Institute, and Drs Steven James and Geraud Nangue Tasse from CSAM. Notably, James was also a Google PhD Fellow in 2018 — highlighting the fellowship’s lasting and recurring impact on machine learning research in Africa.
Bester’s research focuses on the fundamental theory underlying modern AI systems, with a strong emphasis on understanding why algorithms work, when they fail, and how they can be designed more reliably.
His work bridges neural and symbolic approaches to AI, aiming to unify low-level learning with high-level reasoning. He introduced Counting Reward Machines, the first Turing-complete framework for state-machine-based reward modelling, establishing a new theoretical foundation for optimal behaviour in reinforcement learning tasks requiring complex temporal reasoning.
Building on this, his PhD research explores optimal knowledge transfer between subtasks, principled automated reward design, and the synthesis of task structure from natural language—developing rigorous, general-purpose algorithms to improve the reliability, interpretability, and scalability of reinforcement learning.
Global recognition, local impact
The success of Rees and Bester underscores the growing international recognition of Wits University as a hub for world-class AI and machine learning research. Their work exemplifies how foundational theory and applied research can work hand in hand to address pressing societal challenges, from healthcare delivery to the development of more robust and trustworthy intelligent systems.
The Google PhD Fellowship not only supports these students individually, but also strengthens the broader research ecosystem at Wits and across Africa—enabling talented researchers to pursue ambitious ideas, build global collaborations, and contribute knowledge with lasting impact.