About the position
AI Consultant / Specialist
Location: Bellville-South, Cape Town, Western Cape, South Africa (Hybrid model)
Duration: The 6-month contract (possibility of an extension)
Purpose of the Role
The AI Consultant / Specialist is responsible for providing applied AI expertise to the COD Task Team and assist in determining which AI technologies and services are feasible to implement, including machine learning, generative AI, and large language models. This includes technical leadership on AI Sandbox design and deployment, guidance on infrastructure requirements, operating model design, and governance document authoring.
The consultant is not responsible for infrastructure management but will support the Task Team in engagements with the Infrastructure and Operations Team and defining the COD infrastructure requirements in consultation with the Infrastructure and Operations Team.
Key Duties and Responsibilities
Technical Leadership: AI Platform Design, Implementation, and Advisory
- Lead the technical design of the COD AI Sandbox environment, including platform architecture, model selection (across ML, GenAI, LLM technologies), inference infrastructure, and user access model.
- Evaluate and recommend deployment options: on-premises (University HPC), cloud-hosted, or hybrid - taking into account ICS HPC capacity, POPIA obligations, and cost constraints.
- Lead or support the deployment, configuration, and testing of a secure and sustainable AI Sandbox, including hands-on implementation where required.
- Assess open-source and vendor platforms for suitability across the COD's technology scope.
- Recommend and implement appropriate AI model and AI solution development practices for lifecycle management, including versioning, monitoring, and update processes.
- Ensure alignment with enterprise artifacts, principles, and standards as well as institutional, IT and digital strategies.
- Collaborate with enterprise and solutions architects.
Infrastructure Assessment and Requirements
- Develop an understanding of the university's existing HPC and GPU infrastructure: node inventory, VRAM capacity, utilisation, cooling, and power adequacy for sustained AI workloads.
- Analyse, document and communicate requirements for COD AI workloads considering available HPC capacity and identify gaps.
- Engage with various stakeholders regarding infrastructure requirements.
- Identify and document feasible infrastructure alternatives if on-premises capacity is insufficient.
- Collaborate with e-Research Team, Infrastructure and Operations Team and Network Teams.
Governance and Policy Documentation
- Assist and provide advice for with the drafting of foundational governance documents for the COD, including:
- Acceptable Use Policy (AUP) for the AI Sandbox
- AI Governance Document
- Relevant AI Security Artifacts
- COD Operating Model
- COD Service Catalogue
- All governance documents must be aligned with POPIA requirements, university's institutional data classification framework, relevant university policies and guidelines, and relevant AI ethics and governance principles.
- Collaborate with eResearch Team and Governance, Risk and Compliance Team.
Service Design and Roadmap
- Assess feasible AI services that the COD can realistically offer given university's infrastructure, budget, and staffing constraints.
- Assist with producing a phased COD implementation and resourcing roadmap.
- Ensure COD AI services are distinct from eResearch AI services.
Stakeholder Engagement and Knowledge Transfer
- Participate actively in Task Team meetings and working sessions, providing technical input in accessible language for non-technical stakeholders.
- Engage with relevant ICS teams and business stakeholders as required to map service boundaries and assist with negotiating resource agreements.
- Document all design decisions, technical rationale, and configuration specifications to ensure institutional knowledge is retained beyond the contract period.
- Collaborate with relevant ICS team members as part of skills and knowledge transfer if required.
Ad-Hoc AI Support, Advisory, and Incident Response
Secondary to COD establishment work, the consultant will be available to COD Task Team members for ad-hoc AI-related support and advisory services. This support is scoped to the Task Team only and does not extend to the broader ICS department or the University at this stage. COD establishment deliverables retain priority; ad-hoc requests are accommodated within available capacity and must not displace scheduled Task Team work without agreement from the Task Team Lead or ICS Leadership Team. These requests may include:
- AI tool selection guidance: advise COD Task Team members on the selection of AI tools and platforms for specific use cases, including assessment of fitness-for-purpose, licensing implications, data handling risks under POPIA, and alignment with the University's acceptable use framework, polices and guidelines.
- AI incident triage and response: serve as a first point of escalation for AI-related incidents within ICS - including unexpected model outputs, data exposure concerns arising from AI tool use, and misuse or policy-breach situations - and advise on appropriate remediation or referral to GRC.
- Horizon-scanning and awareness: flag emerging AI tools, risks, or developments relevant to COD - including new vendor offerings, changes to licensing or data-use terms of tools already in use, and significant model capability changes that may affect AI governance.
- Ad-hoc proof-of-concept work: on request from the Task Team, scope and execute small-scale AI proof-of-concept investigations to inform COD decision-making.
Preferred Duties
- Vibe-coding (AI generated applications) platform and tool selection within the context of COD in collaboration with relevant ICS stakeholders and architects.
- Develop vibe-coding (AI generated applications) guidelines and processes to support business users using AI tools to generate applications - including prompt engineering, output validation, security review of AI-generated code, and responsible use practices within the context of COD.
Minimum Qualifications
- Bachelor's degree or equivalent in Computer Science, Information Technology, Data Science, Artificial Intelligence, or a related field.
- Relevant professional certifications in AI/ML, cloud platforms, or MLOps are advantageous.
Experience Requirements
Required Experience
- Minimum 5-10 years of demonstrated experience in applied AI or machine learning, combining hands-on technical implementation with advisory or specialist input - including direct experience deploying or configuring AI/ML systems.
- Demonstrated experience designing and deploying AI/ML platforms or inference environments - including machine learning pipelines, GenAI applications, or LLM-based systems - in production or near-production settings (self-hosted or cloud-hosted).
- Experience working with GPU infrastructure for AI workloads, including understanding of compute resource allocation and scheduling.
- Experience producing governance, policy, or technical documentation in a structured organisational context.
Advantageous Experience
- Experience in a higher education, research, or public sector environment.
- Familiarity with South African data protection law (POPIA) and its implications for AI system design.
- Experience with HPC infrastructure.
- Prior involvement in AI governance, responsible AI, or AI ethics frameworks.
- Experience with open-source AI/ML platforms and frameworks.
- Experience producing architecture documentation.
- Familiarity with MLOps platforms
Behavioural Competencies
- Ability to work effectively within a small, focused task team with limited resources and significant ambiguity.
- Strong written communication: able to produce governance documents and technical specifications that are clear, precise, and usable by non-technical readers.
- Pragmatic problem-solving: able to identify realistic, achievable solutions within institutional constraints rather than ideal-state recommendations.
- Professional integrity: able to provide honest, evidence-based assessments of what is and is not feasible - including communicating risks and limitations clearly to leadership.
Working Arrangements and Governance
- The AI Consultant reports to the COD Task Team Lead for all task team deliverables and activities.
- Progress reporting: the consultant will participate in regular Task Team meetings and provide written status updates to the Task Team Lead at agreed intervals.
- All work products produced under this appointment remain the intellectual property of the University / ICS. The consultant must not share, publish, or reuse institutional data or documentation outside of the engagement without written authorisation.
- Confidentiality: the consultant will oversee all the university's institutional data, systems access, and governance documents in accordance with POPIA and university's information security policies.
Desired Skills:
- AI/ML
- MLOps
- GenAI
- HPC infrastructure
- AI governance
