Surtee T1
1The University of the Witwatersrand, Johannesburg, South Africa
Biography:
Dr Taariq Surtee is the Head of eResearch at the University of the Witwatersrand, Johannesburg (Wits University), a position he has held since January 2018. In this role he leads the University's strategy for digital research infrastructure, overseeing the Wits Core Cluster (~1,300 cores, 1.3 PB of storage), research data management (RDM), institutional data repositories (Figshare, DSpace), high-performance computing user support, and emerging technology programmes including quantum computing (WitsQ / SA QuTI). .
A self-described "pracademic," Taariq bridges academic research and professional practice. He holds a PhD in Financial Mathematics (optimisation), an MBA, and a BSc Honours in Applied Mathematics, all from Wits. His doctoral research on performance evaluation and portfolio optimisation honed the analytical skills he now applies to optimising shared research computing infrastructure. His publications span financial mathematics (a novel approach to modern portfolio theory, *Borsa Istanbul Review*, 72 citations), Islamic finance (*Journal of Economic Cooperation & Development*), quantum computing education (SAIP 2025), and digital humanities (*DHASA Journal*). He is an accomplished ICT programme manager and is recognised by the International Professional Recognition Council as a Research Management Professional, actively supporting its STARS professionalism advocacy programme.
Before leading eResearch, Taariq served as Senior Manager of ICT Programmes at Wits, managing university-wide technology infrastructure projects including network backbone upgrades, enterprise information systems, and computer data centre and laboratory builds. Earlier roles as Business Intelligence Engineer at Wits, Acting Director in the Gauteng Provincial Government's Programme Management Unit, and Data Scientist at Liberty Life — where he designed statistical models for customer behaviour and market segmentation — gave him deep experience in data science, statistical modelling, project governance, and large-scale technology delivery. This diverse background directly informs his current work on making university HPC competitive with commercial cloud offerings.
Taariq is the co-founder and Chair of the Organising Committee for the African Data Science Conference (ADSC). The 2026 conference will be held 24–26 June at Wits under the theme "African Data Science for Africa." ADSC is a growing continental forum for data science, AI, machine learning, and analytics with an emphasis on worldwide contexts. He also serves on the programme and scientific committees. The conference reflects his broader mission: building Africa's research technology capacity so African institutions are not just consumers of global digital infrastructure but active shapers of it — a mission aligned with his AI-managed HPC presentation in this conference.
Taariq serves on Wits University Senates, Council, HR Council, Finance Committee, and Tender Committee. He actively engages with national and international bodies, including SARIMA, CARTA, HEITSA, CSIR, IBM, Microsoft, and Amazon, and shares his expertise with sister universities across South Africa. He teaches quantum computing and research data management whenever his schedule allows mentors emerging researchers and HPC student teams for the CHPC Student Cluster Competition and is passionate about using technology to give researchers a globally competitive edge.
Abstract:
University on-campus HPC faces an existential question: why should researchers use our clusters when hyperscalers offer elastic, fully-managed cloud compute on demand? They shouldn't — unless we rethink how we run our facilities. This presentation argues that AI is not just a workload to run on HPC, but a tool to *manage* it.
At the Wits Core Cluster (~1,300 cores, 1.3 PB), we are exploring AI-driven approaches that enable a team to compete with the cloud on five fronts. First, affordability: predictive scheduling and ML-based resource optimisation push cluster utilisation well beyond academic averages, lowering cost-per-core-hour below what cloud can match at sustained workloads. Second, staff retention: the chronic shortage and poaching of systems engineers is the biggest threat to on-campus HPC, and AI-assisted monitoring, auto-remediation, and intelligent support reduce the operational load, making roles more strategic and less about firefighting. Third, security: on-campus infrastructure keeps sensitive data under institutional and legislative control (e.g. POPIA), and AI-augmented threat detection strengthens that advantage rather than ceding it to third-party tenancies. Fourth, a hybrid model: an AI-managed on-campus core handles the baseline, with policy-driven bursting to commercial cloud for peak demand or specialised GPU workloads. Fifth, system design: we share lessons from architecting a platform whose management layer — scheduling, storage tiering, onboarding, support — is AI-augmented from the ground up.
Universities will never outspend the cloud giants — but with AI in the management plane, we can out-think them, keep our talent, protect our data, and offer an affordable, sovereign alternative.
