Prompt Engineer

About the role

We are looking for a talented and detail‑oriented Prompt Engineer to join our growing AI team in Dubai. In this role you will craft, evaluate and optimise prompts that drive the behaviour of large‑language‑model‑powered applications and intelligent agents.

Key responsibilities

• Design, test and systematically optimise prompts for LLM‑powered applications and AI agent pipelines.
• Create and continuously refine system prompts, guardrails, evaluation prompts and multi‑agent interaction frameworks.
• Collaborate with AI backend engineers to integrate prompts into production APIs and agent workflows.
• Develop comprehensive prompt evaluation datasets, scoring methodologies and automated testing pipelines.
• Experiment with Retrieval‑Augmented Generation (RAG) pipelines, memory systems, tool‑calling mechanisms and agent orchestration strategies.
• Analyse agent failures and iterate on prompt designs using structured evaluation frameworks and real‑world feedback.
• Partner with product and business stakeholders to translate functional requirements into precise AI behaviours.
• Document prompting strategies, best practices and maintain a library of reusable prompt templates.

Required profile

• 2+ years of hands‑on experience with large language models and generative AI in production.
• Strong command of prompt‑engineering techniques including reasoning, information extraction, summarisation and agentic workflow design.
• Proven experience with synthetic dataset generation and AI evaluation methodologies.
• Familiarity with Model Context Protocols, AI gateways and model routing strategies.
• Solid understanding of AI safety principles, prompt‑injection risks and secure agent design.
• Experience building and evaluating AI agents using tool‑calling, RAG architectures and memory systems.

Required skills

• Python programming.
• LangChain, LangGraph, LlamaIndex (or equivalents).
• Retrieval‑Augmented Generation (RAG) pipelines.
• AI monitoring and evaluation tools such as LangSmith, DeepEval, Ragas.
• JSON schema design for LLM responses.
• Model Context Protocols (MCP) and AI gateway integration.
• Prompt‑injection mitigation and hallucination reduction techniques.