AI Prompt Engineer

Role overview

A Prompt Systems Engineer position focused on designing, testing, and continuously improving the prompts and system instructions that power AI agents used across higher education contexts such as enrollment, learning, and student support. The work sits at the intersection of AI capability and educational effectiveness, partnering with product managers, learning designers, engineers, and university stakeholders to translate pedagogical and operational goals into reliable agent behavior. It is a hands-on, high-impact role with direct ownership over prompt quality, evaluation practice, and ongoing improvement.

Responsibilities

- Write, iterate, and maintain system prompts and instruction sets for AI agents used in enrollment, learning, and retention workflows - Translate complex educational and operational requirements into clear, dependable agent behavior, including persona, tone, scope constraints, fallback handling, and multi-turn conversation logic - Design prompt architectures for multi-step and chained agent flows, including retrieval-augmented patterns over program-specific knowledge bases - Build and run evaluation frameworks that measure accuracy, tone, hallucination rate, task completion, and alignment with rubric-based learning objectives - Use LLM observability tooling to monitor production performance, detect regressions, and prioritize prompt improvements - Design red-teaming and adversarial testing protocols to surface edge cases before agents reach students - Establish prompt versioning practices and curate a library of tested, reusable prompt components - Partner with internal teams and external stakeholders to convert learning objectives, operational flows, and rubric assessments into prompt-level instructions - Stay current with capability changes across major model providers and surface opportunities or risks

Requirements

- 2+ years designing and iterating on prompts for LLM-powered applications in a production environment - Deep familiarity with prompt patterns such as few-shot examples, chain-of-thought reasoning, system versus user roles, output formatting constraints, tool-use prompting, and retrieval-augmented generation - Strong written communication skills, with the ability to write clearly, precisely, and persuasively - Experience building prompt evaluation frameworks: defining metrics, writing test cases, and interpreting results to drive improvements - Comfort working across disciplines, from discussing context windows with engineers to discussing pedagogy with faculty - Experience with LLM observability platforms such as Langfuse or Weights & Biases - Familiarity with Learning Tools Interoperability and how AI features integrate with learning management systems - Experience with voice agents or multimodal prompting is welcome

Eligibility

- The position is based in South Africa. Applicants must be legally authorized to live and work in South Africa and must maintain residency there throughout employment