Agentic AI Architect: Orchestrating Safe, Scalable Agents

Mondia Media; Dubai; On-site; AED 300,000 - 520,000; Full time

Job summary

Mondia Media seeks an experienced software engineer to drive AI-powered automation across the organization. You will scout emerging AI technologies, pilot new tools, and coordinate cross‑departmental adoption, enhancing our Knowledge Hub for maximum AI utility.

You will design agentic architectures, multi‑step planners, and memory systems, integrating LLMs with external tools and enterprise systems. Build reliable pipelines, evaluate performance, and implement guardrails with human‑in‑the‑loop

Qualifications

• 5+ years of software engineering experience, with 1–2 years focused on LLM-based systems.
• Hands-on experience building agentic systems, including orchestration, tool use, and state management.
• Strong proficiency in Python; familiarity with TypeScript is a plus.
• Deep understanding of LLM APIs (Claude, OpenAI, etc.), including function calling and streaming.
• Experience with at least one agent framework (LangGraph, CrewAI, AutoGen, or similar).

Responsibilities

• Scout emerging AI technologies and pilot new tools across departments.
• Design agentic architectures, multi-step planners, ReAct loops, and memory systems.
• Build reliable agent pipelines with error handling and human-in-the-loop controls.
• Develop evaluation frameworks for reliability, cost, and latency in production.
• Collaborate with product and engineering to identify automation opportunities.

Skills

Software engineering

LLM-based systems

Agent orchestration

State management

Python

CI/CD

Tools

LangGraph

CrewAI

AutoGen

Claude

OpenAI

Pinecone

Weaviate

pgvector

Job description

Mondia Media seeks an experienced software engineer to drive AI-powered automation across the organization. You will scout emerging AI technologies, pilot new tools, and coordinate cross‑departmental adoption, enhancing our Knowledge Hub for maximum AI utility.

You will design agentic architectures, multi‑step planners, and memory systems, integrating LLMs with external tools and enterprise systems. Build reliable pipelines, evaluate performance, and implement guardrails with human‑in‑the‑loop