Senior AI Platform Engineer (Cloud) - Sandton at Absa South Africa

Job Summary

• Absa Group’s Chief Data Analytics and Applied AI Office (CDAIO) requires a technically exceptional and commercially grounded AI Platform Engineer (Cloud) to design, build, operate, and continuously optimise the multi-cloud AI infrastructure that powers the bank's enterprise AI capability.
• The AI capability must enable the CDAIO to fulfil its mandate as steward of the bank’s AI capabilities through the end-to-end delivery of the AI platform enablement, governance and acceptable use in service of the bank’s strategic and commercial objectives.
• This role is the engineering backbone of a platform that supports various live AI projects across four business units (CIB, PPB, BB, and AR) and ten countries. This role demands deep technical mastery in cloud AI infrastructure, AI FinOps, zero-trust security architecture, agentic AI infrastructure, and platform observability, combined with the commercial fluency to govern AI compute costs at enterprise scale and communicate trade-offs to senior business and finance stakeholders.
• The role includes but not limited to applying critical thinking, design thinking, and problem-solving skills in an agile team environment to solve complex platform engineering challenges, delivering high-quality solutions at optimal cost to serve, in full compliance with Absa's Enterprise-Wide Risk Management Framework, Group Architecture standards, and AI Responsible Use Policy.
• The successful candidate carries full accountability for building high-performing, scalable, enterprise-grade Platform services. As well as build capability in others to do the same.

Job Description

KEY FOCUS AREAS 

• AI Platform Engineering and Architecture: Design and operation of enterprise-grade, multi-cloud AI platform infrastructure supporting bank-wide AI delivery at scale across the AI platform stack (AWS Bedrock, Databricks AI, Microsoft Azure AI Foundry, Hugging Face, and GPU clusters). 
• AI FinOps and Compute Cost Governance: Full accountability for AI compute cost models, chargeback and showback frameworks, provisioned throughput optimisation, and monthly cost-per-use-case reporting to Group Finance across all four business units. 
• Platform Observability and SLA Engineering: AI-specific service reliability standards, observability tooling, and incident management for production AI workloads serving 43 live projects across ten countries. 
• AI Security Architecture and Zero Trust: Zero-trust security design, OAuth / OIDC integration, prompt injection controls, and data residency compliance protecting Absa's AI platform across the different country jurisdictions. 
• Agentic AI Infrastructure: Design and operation of the infrastructure layer enabling multi-agent AI systems, autonomous workflows, tool-calling architectures, and agent orchestration at enterprise scale. 

ACCOUNTABILITIES 

 Platform Engineering and Architecture 

• Lead the design, deployment, and continuous optimisation of Absa's multi-cloud AI platform stack: AWS Bedrock, Databricks AI, Microsoft Azure AI Foundry, Hugging Face Model Hub, and on-demand GPU clusters. 
• Architect scalable, resilient, and reusable platform components including AI Gateway configuration, model serving infrastructure, vector database deployments, and data pipeline integration to support bank-wide AI delivery. 
• Define and maintain infrastructure-as-code (IaC) standards (e.g. using Terraform or Pulumi), enabling repeatable, auditable multi-cloud AI deployments across Absa's operating territories (10 countries). 
• Lead the design and operation of agentic AI infrastructure: orchestration runtime environments (e.g. Microsoft Foundry Agent Service, AWS Bedrock Agents), tool-calling schemas, agent memory and state management patterns, and multi-agent communication protocols. 
• Develop and enforce cloud-agnostic model serving patterns to reduce platform lock-in and ensure workload portability across the CDAIO's multi-vendor stack. 
• Identify and select appropriate internal and external technologies to deliver AI platform services; apply excellent judgement in continuously improving platform engineering practices. 
• Take full accountability for end-to-end platform quality, completeness, and user experience across the development, deployment, and operational lifecycle. 
• Positively contribute to the design and evolution of Group Architecture, infrastructure standards, and AI platform governance frameworks 

 AI FinOps and Compute Cost Governance 

• Own the AI compute cost model for the CDAIO, including chargeback and showback frameworks for Databricks DBU consumption, AWS Bedrock token-based pricing, Azure AI Foundry provisioned throughput units, and GPU cluster utilisation across all four business units. 
• Design and maintain FinOps dashboards and cost attribution reports using AWS Cost Explorer, Databricks System Tables cost analytics, and Azure OpenAI utilisation tooling — providing monthly cost-per-use-case reporting to Group Finance and the CDAIO COO. 
• Evaluate and manage provisioned throughput versus on-demand consumption trade-offs for production AI workloads, presenting optimisation recommendations to the CDAIO and BU technology leads. 
• Identify and execute AI compute cost optimisation opportunities: workload scheduling, spot instance strategies for training workloads, model distillation to reduce inference cost, and right-sizing of GPU clusters. 
• Create business cases and solution specifications for AI platform investments and governance processes, including CTO and architecture approvals. 
• Collaborate with the FinOps capability within the CDAIO COO to align AI platform costs to agreed budget envelopes and ensure spend anomalies are detected and escalated proactively  

Platform Observability and SLA Engineering 

• Define, implement, and own AI-specific SLAs and OLAs covering inference latency, platform availability, token throughput, API gateway response times, and model serving reliability, with explicit targets agreed with each business unit technology lead. 
• Implement and maintain AI platform observability tooling (e.g. Prometheus, Grafana, Datadog, Databricks Lakehouse Monitoring, or equivalent) providing real-time visibility of platform health, model drift alerts, and capacity utilisation. 
• Design and operate incident management processes for AI platform failures: on-call runbooks, escalation paths, post-incident reviews, and root-cause remediation, ensuring minimal disruption to live AI projects across Absa's footprint. 
• Lead service improvement initiatives, translating performance data into platform enhancement programmes and continuously reducing mean time to recovery (MTTR) across the platform estate. 
• Own the release and change management process for AI platform components, including change governance, cutover management, and operational readiness sign-off in alignment with Absa's Group Technology change framework. 
• Use production performance monitoring and customer data to inform technical design and implementation decisions; leverage systems and processes to measure, monitor, and manage platform performance bank-wide 

 AI Security Architecture and Zero Trust 

• Design and implement zero-trust security architecture for AI platform APIs and services such as OAuth 2.0 / OIDC integration, JWT/JWE/JWS token management, role-based access control (RBAC), and attribute-based access control (ABAC) for AI workloads. 
• Implement prompt injection prevention, output filtering, and data exfiltration controls at the AI Gateway layer, protecting data confidentiality for all LLM and agentic AI interactions across business units. 
• Design and enforce data residency and sovereignty controls for AI platform deployments across Absa's operating countries, ensuring compliance with country-specific data localisation requirements and cross-border data transfer restrictions. 
• Conduct and maintain AI-specific threat models in collaboration with the Chief Information Security Office, covering third-party AI vendor risks (Databricks, AWS, Microsoft, Hugging Face), model supply chain integrity, and adversarial ML attack vectors. 
• Apply and maintain all Group risk, governance, compliance, and regulatory standards and frameworks; hold accountability for all risk associated with AI platform engineering decision-making. 
• Update, develop, and maintain all platform documentation in accordance with organisational technical standards and risk and governance frameworks. 

 People, Capability and Agile Delivery 

• Lead and develop a team of AI Platform Engineers, establishing clear performance objectives, providing regular coaching and feedback, and building a high-performance, self-directed squad aligned to agile delivery practices. 
• Cascade platform direction across the team; ensure alignment on platform strategy, performance objectives, and delivery priorities. Assume end-to-end accountability for the right people in the right teams to deliver the platform strategy. 
• Leverage coaching techniques across all squad-related activity to drive higher-quality design and deployment of AI platform services. 
• Maintain comprehensive technical documentation, architectural decision records (ADRs), and operational runbooks for all platform components, ensuring service continuity is independent of individual staffing changes and contractor dependencies are actively mitigated. 
• Conduct peer reviews, testing, and problem-solving within and across the broader CDAIO engineering community; identify and develop needed skills in self and others. 
• Support the AI Embedment and Training capability in developing platform onboarding materials and self-service guides to accelerate business unit adoption of AI platform services. 
• Proactively lead agile practices, remove barriers to success, and ensure seamless delivery in a continuously changing environment. 

QUALIFICATIONS AND EXPERIENCE 

 Education/ Qualification: 

• Postgraduate degree in a quantitative discipline such as Computer Science, Data Science, Mathematics, Statistics, Engineering, or equivalent ((Masters-essential or PhD-advantageous). 

Certification in: 

• Cloud - AWS Solutions Architect Professional, AWS Machine Learning Specialty, or Microsoft Azure AI Engineer Associate). 
• FinOps - FinOps Foundation Certified Practitioner (FOCP) or equivalent AI cost governance credential. 
• Security Certification - Certified Cloud Security Professional (CCSP) or AWS Security Specialty. 
• IaC Certification - HashiCorp Terraform Associate or equivalent infrastructure-as-code credential. 

 Work Experience: 

• 5-8 years of progressive leadership experience in Cloud AI Platform Engineering, with production experience managing multi-cloud AI platform stacks across at least two of: AWS Bedrock/SageMaker, Databricks AI, Microsoft Azure AI Foundry, or Hugging Face enterprise deployments. 

2–3-year experience in the following: 

• AI FinOps and Cost Governance: Demonstrated ownership of AI compute cost models and FinOps reporting in a multi-BU or multi-cloud environment, with evidence of cost optimisation outcomes. 
• AI Security Architecture: Designing and implementing zero-trust AI security (OAuth/OIDC, JWT), prompt injection controls, data residency compliance in a regulated environment. 
• Agentic AI Infrastructure: Production design of agent orchestration infrastructure (such as LangGraph, AutoGen, Foundry Agent Service, Bedrock Agents), tool-calling APIs, and agent state management. 
• Platform Observability: Operating AI-specific observability tooling for inference latency, drift alerting, and capacity management (such as Prometheus, Grafana, Datadog, or Lakehouse Monitoring). 
• Infrastructure-as-Code: Terraform, Pulumi, or equivalent for multi-cloud, multi-region AI infrastructure deployments; CI/CD pipeline design for platform components. 
• Regulated Industry: AI platform engineering in financial services or a similarly regulated sector with model risk governance and change management obligations.  
• Regulated Industry: AI platform engineering in financial services or a similarly regulated sector with model risk governance and change management obligations 

Advantageous: 

• People leadership: Leading or mentoring a team of platform or infrastructure engineers in an agile delivery environment. 
• Pan-African Deployments: Delivering AI platform services across multiple African jurisdictions with awareness of data localisation and cross-border data transfer requirements. 

Knowledge and Skills:

• Multi-Cloud AI Platform Architecture: Expert design and operation of AWS Bedrock, Databricks AI, Azure AI Foundry, and Hugging Face in enterprise production environments across multiple business units and geographies.
• Agentic AI Infrastructure: Practical production knowledge of agent orchestration frameworks (LangGraph, AutoGen, Foundry Agent Service, Bedrock Agents), tool-calling API design, agent memory architecture, and multi-agent coordination patterns.
• AI FinOps and Cost Management: Chargeback and showback model design; DBU and token cost attribution; provisioned throughput versus on-demand optimisation; GPU cluster cost management; spend anomaly detection and FinOps dashboarding.
• AI Security and Zero Trust: OAuth 2.0, OIDC, JWT/JWE/JWS; RBAC and ABAC for AI workloads; prompt injection prevention; data exfiltration controls at the Gateway layer; AI threat modelling and data residency compliance.
• Infrastructure-as-Code: Terraform, Pulumi, or AWS CDK for multi-cloud AI infrastructure; CI/CD pipeline design for platform components; container orchestration using Docker, Kubernetes, and Helm.
• Platform Observability: Prometheus, Grafana, Datadog, OpenTelemetry, and Databricks Lakehouse Monitoring; custom metric design for AI workload health including inference latency, token throughput, and model drift.
• Cloud-Agnostic Model Serving: ONNX, BentoML, Triton Inference Server; containerised model deployment patterns for portability across AWS, Azure, and Databricks environments.
• MLOps Tooling: Working knowledge of MLflow, Kubeflow, Airflow, and CI/CD for ML, sufficient to collaborate effectively with AI Solution Engineers on model deployment and lifecycle management
• GPU and HPC Architecture: On-demand GPU cluster management; spot instance strategies; high-performance compute cost optimisation for large-scale model training and fine-tuning workloads.
• Enterprise Risk and Governance: Absa Enterprise Wide Risk Management Framework; Group Architecture standards; AI Responsible Use Policy; POPIA; country-specific data localisation requirements across Absa's ten operating countries.
• Agile Delivery: Sprint planning, backlog management, and continuous delivery practices in a self-directed squad environment; experience removing delivery barriers in a fast-moving, multi-stakeholder context.

Education

• Bachelor's Degree: Information Technology

End Date: October 16, 2026