AI Engineer

Description

Purpose of the Role

The AI Engineer is responsible for designing, building, testing, deploying and supporting secure AI-enabled applications, intelligent agents, workflow automations and enterprise integrations. The role translates approved business requirements and future-state processes into maintainable and measurable solutions that improve service delivery, productivity, operational control and client experience.

The AI Engineer works closely with the Business Analyst, Process Engineer, Software Engineering, Enterprise Solutions, Data & AI, Information Security, Project Management, business stakeholders and external delivery partners.

The role applies a governed and human-controlled engineering approach. AI-assisted outputs must be reviewed, tested and approved before deployment. Legal, statutory, financial and compliance-related decisions must remain deterministic, auditable or subject to appropriate human approval.

Duties and ResponsibilitiesStrategy & Planning

• Review approved business requirements, process maps, business rules, data definitions, controls and acceptance criteria with the Business Analyst and Process Engineer.
• Identify unclear requirements, conflicting rules, data gaps, security considerations and non-functional requirements before development begins.
• Translate approved requirements into engineering-ready work items with clearly defined inputs, outputs, constraints and acceptance tests.
• Determine whether a requirement is best addressed through deterministic software, workflow automation, agentic AI, retrieval, low-code configuration or a combination of solutions.
• Assess and recommend appropriate AI tools, platforms and models based on architecture, security, licensing, cost and business requirements.
• Define measurable success criteria relating to accuracy, reliability, workflow completion, security, operational performance and business value.

Development & Deployment

• Design and develop AI-enabled applications, agents, agent flows, APIs, integrations, validation services, data components and user interfaces.
• Build governed enterprise agents and workflows using Microsoft Copilot Studio, Power Automate, Azure AI services and approved Microsoft 365 technologies.
• Use approved AI-assisted development tools, including Claude Code and GitHub Copilot, to support software development, testing, documentation and refactoring.
• Develop retrieval-augmented generation and document-intelligence solutions using approved, authoritative and access-controlled information sources.
• Build controlled integrations with platforms such as Patricia, Aderant, NetDocuments, Sage, October People and Microsoft 365.
• Implement appropriate human approval, access control, exception handling, audit trails and escalation mechanisms within AI-enabled solutions.
• Ensure AI-generated software artefacts pass human review, automated testing, security checks and approved repository and deployment controls.
• Maintain appropriate versioning, testing, deployment, configuration and rollback processes for code, prompts, agents, models and connectors.

Operational Management

• Develop and maintain evaluation datasets and test suites covering accuracy, grounding, workflow completion, tool use, security boundaries and unacceptable behaviour.
• Conduct unit, integration, regression and end-to-end testing across normal, ambiguous, adversarial and failure scenarios.
• Monitor solution availability, latency, error rates, tool calls, workflow outcomes, human overrides and AI consumption costs.
• Investigate incidents, failed workflows and degraded outputs, conduct root-cause analysis and implement corrective controls.
• Apply information security, confidentiality, legal privilege, POPIA, records-management and responsible AI requirements throughout the solution lifecycle.
• Implement role-based access, least-privilege permissions, approval gates, audit logs, content safeguards and emergency disablement controls.
• Maintain current technical documentation, operational runbooks, architecture records, deployment information and escalation procedures.
• Measure delivered solutions against agreed business outcomes, including turnaround time, manual effort, rework, errors, service quality and control performance.
• Demonstrate working solutions, incorporate stakeholder feedback and support user training, adoption and knowledge transfer.
• Develop reusable prompts, connectors, evaluation assets, coding standards and engineering patterns to improve consistency and reduce future delivery effort.

Key improvements

The original responsibilities have been grouped into Strategy & Planning, Development & Deployment, and Operational Management, matching the structure of your Business Analyst example. The wording has also been condensed without removing the role’s core focus on agentic AI, enterprise integration, responsible AI, testing, security and operational reliability.

Requirements

QualificationsEducation

• Relevant Bachelor’s degree or diploma in Computer Science, Software Engineering, Data Science, Information Systems, Artificial Intelligence or a related discipline.
• Relevant certifications in Microsoft Azure AI, Power Platform, Copilot Studio, software development, cloud, data or security would be advantageous.

Experience

• Typically 4-7+ years’ experience in software engineering, integration, automation, data engineering or AI solution delivery, including practical Generative AI or Agentic AI experience.
• Demonstrated experience building and supporting production software, APIs, integrations, automated testing and controlled deployments.
• Practical experience designing and implementing AI agents, workflow orchestration, tool use, human approvals, grounding, evaluation and monitoring.
• Experience using AI-assisted development tools within a controlled software development and DevOps environment.
• Experience integrating cloud and enterprise platforms through APIs, connectors, databases, events and automated workflows.
• Experience within legal, financial, professional services or another regulated environment handling confidential information would be advantageous.

Behavioural Competencies

• Engineering Judgement: Balances AI capabilities with security, maintainability, cost, business requirements and operational risk.
• Analytical Thinking: Breaks complex requirements into clear, logical and testable components.
• Curiosity & Continuous Learning: Keeps abreast of evolving AI technologies and assesses new tools pragmatically.
• Communication: Communicates technical concepts, AI limitations, risks and design decisions clearly to technical and non-technical stakeholders.
• Collaboration: Works effectively across business, technology, legal, security, data and external delivery teams.
• Constructive Challenge: Confidently challenges unclear requirements, inappropriate automation and potential risks.
• Attention to Detail: Maintains accurate configurations, testing, permissions, documentation and deployment controls.
• Commercial Awareness: Considers licensing, infrastructure, AI consumption and support costs against expected business value.
• Integrity & Confidentiality: Protects client, employee and firm information and consistently applies responsible AI principles.
• Accountability: Takes ownership of solution quality, identifies limitations early and drives corrective action where required.

Technical Competencies

• AI-Assisted Development: Experience with Claude Code, GitHub Copilot or other approved coding assistants for development, code review, testing, documentation and refactoring.
• Agent & Automation Platforms: Strong working knowledge of Microsoft Copilot Studio, Power Automate, Azure AI services and Azure OpenAI.
• Software Engineering: Experience with technologies such as C#, .NET, Python, TypeScript or JavaScript, REST APIs, JSON, SQL, Git, automated testing and CI/CD.
• Microsoft Technology Stack: Knowledge of Azure, Azure DevOps, Microsoft 365, Microsoft Graph, Power Platform, Key Vault and Application Insights.
• Agent Engineering: Knowledge of prompt and context engineering, tool/function calling, agent state, routing, approval gates, exception handling and agentic workflows.
• Retrieval & Knowledge: Understanding of RAG, embeddings, document ingestion, chunking, metadata, permissions, grounding and source citation.
• AI Evaluation: Experience with evaluation datasets, groundedness and accuracy testing, regression testing, adversarial testing and human review.
• LLMOps & Observability: Understanding of versioning, deployment, rollback, telemetry, tracing, logging, cost monitoring and operational runbooks.
• Responsible AI & Security: Strong understanding of POPIA, confidentiality, legal privilege, data minimisation, access control, auditability, threat modelling and secure software development practices.
• Enterprise Integration: Experience integrating business systems and enterprise platforms through secure APIs, connectors, services and other approved integration methods.