Senior AI Engineer, Agentic Workflows & LLM Pipelines

Halian | Managed Services, Recruitment Agency & Contract Staffing; Abu Dhabi; On-site; AED 300,000 - 550,000; Full time

Job summary

Halian seeks an AI Engineer to join a leading financial institution's AI Centre of Excellence in Abu Dhabi. You will design, build, and test LangGraph agent graphs, LLM extraction pipelines, and prompt management workflows to power intelligent automation in banking operations.

Role involves HITL, guardrails, and multi‑band testing, with a dual orchestration stack (Orkes Conductor + LangGraph) and Azure AI Foundry for production‑grade agentic systems.

Qualifications

• 5+ years of Python development in production environments.
• Hands-on with LangGraph, LangChain, or equivalent agent frameworks.
• Practical LLM prompt engineering with structured outputs and schema enforcement.
• Familiarity with Azure AI Foundry / OpenAI API or equivalent.
• Understanding of agent design patterns: ReAct, self‑reflection, HITL.

Responsibilities

• Agent graph development for supervisor–worker orchestration and HITL patterns.
• Build and maintain LLM extraction pipelines with structured outputs and per-field scoring.
• Author and govern prompts through the management system; ensure testing and versioning.
• Develop and extend MCP tool servers for new use case requirements.
• Implement AI guardrails: content safety, PII detection, and prompt shielding.
• Validate designs with multi‑band testing and maintain datasets.

Skills

Python

Prompt engineering

REST API

Git & CI/CD

OpenAI / LLM APIs

Education

Bachelor's degree in Computer Science / AI / Software Engineering

Tools

LangGraph

LangChain

Azure AI Foundry

OpenAI API

Job description

Halian seeks an AI Engineer to join a leading financial institution's AI Centre of Excellence in Abu Dhabi. You will design, build, and test LangGraph agent graphs, LLM extraction pipelines, and prompt management workflows to power intelligent automation in banking operations.

Role involves HITL, guardrails, and multi‑band testing, with a dual orchestration stack (Orkes Conductor + LangGraph) and Azure AI Foundry for production‑grade agentic systems.