Artificial Intelligence Engineer

Agentic AI Engineer The OpportunityWe're working with a fast-moving organisation building production-grade agentic AI systems not prototypes, not proof-of-concepts, but reasoning agents that use tools, maintain state, retrieve knowledge, execute multi-step workflows, and know when to escalate to a human.This is a rare chance to own the full lifecycle of agentic AI in production: architecture, orchestration, tool integration, evaluation, safety, observability, and cost optimisation across Azure OpenAI, Anthropic, open models, and provider-native agent SDKs.What You'll DoDesign and operate production agentic AI systems with robust tool-use, memory, and multi-step reasoningSelect and integrate LLMs and agent frameworks (LangGraph, LangChain, LlamaIndex, Semantic Kernel, AutoGen, CrewAI, or custom orchestration), balancing latency, cost, quality and complianceEngineer reliable tool-use patterns: function calling, structured outputs, MCP servers, permissions, retries, sandboxing, and audit trailsBuild retrieval and context-engineering pipelines: RAG, hybrid search, re-ranking, short/long-term memory, and context budgetingOwn agent evaluation and safety: golden datasets, red-teaming, prompt-injection defences, PII controls, and production feedback loopsOptimise for cost and performance: prompt caching, token budgeting, model routing, and batchingDeploy agents as reliable services with CI/CD, canary releases, rollback plans, and provider failoverWhat You'll Bring4+ years in software, ML, or production AI engineering, with proven experience shipping reliable services (not just prototypes)2+ years working hands-on with LLMs prompt/context engineering, structured outputs, function calling, RAG, evaluation, and production monitoringStrong Python skills: async programming, type hints, Pydantic-style validation, API design, and secure, maintainable service architectureDeep knowledge of agentic patterns: ReAct, plan-and-execute, reflection, state graphs, multi-agent orchestration, and human-in-the-loop designExperience with vector databases and retrieval systems (Pinecone, Weaviate, FAISS, Azure AI Search) and evaluation/observability tooling (LangSmith, Braintrust, Application Insights)Understanding of AI security and governance: prompt injection, data exfiltration, sandboxing, and compliance-ready audit trailsNice to have: MCP server authoring, Agent-to-Agent (A2A) protocols, multimodal agents, Arabic/UAE localisation experience, regulated-industry AI governance, and local inference (vLLM/TensorRT-LLM)