AI Engineer - Agentic AI and LLM Systems

HCT Group; Dubai; On-site; AED 300,000 - 520,000; Full time

Benefits offered by this job

UAE employment visa and medical保险

Job summary

HCT Group in Dubai seeks an AI Engineer to design, build and ship agentic AI and LLM systems used in telecoms, government and enterprise contexts. This is an on-site, full-time role with flexible hours and data sovereignty requirements.

You will work with LLM agents, MCP servers, Odoo integrations, and AWS/on-prem GPU deployments, shaping production-grade AI that reads data, drafts proposals and automates workflows.

Qualifications

• 3+ years software engineering experience.
• At least 1 year shipping LLM applications to production.
• Strong Python; TypeScript is a plus.
• Hands-on with LLM APIs (Claude, OpenAI or open models), vector databases and agent frameworks.
• Docker, Linux, CI/CD and cloud (AWS preferred).
• Clear written English and documenting what you build.
• Inference serving with vLLM or TGI, MLOps.
• ERP and Odoo integrations.

Responsibilities

• Build LLM agents with tool use, retrieval (RAG) and memory, connected to Odoo, email, network management and monitoring systems.
• Design MCP servers and APIs that let agents act safely on real systems.
• Build evaluation sets, guardrails and logging so every agent is measurable and auditable.
• Deploy on AWS and on-prem NVIDIA GPU for clients who need data sovereignty.
• Adapt open models (Llama, Qwen, Mistral) for telecom and Arabic/English use cases.
• Work with presales on demos and proofs of concept for operator and government clients.

Skills

Python

TypeScript

LLM APIs

Docker

Linux

CI/CD

AWS

English writing

Odoo integrations

ERP integrations

Tools

vector databases

agent frameworks

Odoo

NVIDIA GPU

MLOps tools

Job description

AI Engineer - Agentic AI and LLM Systems

HCT runs agentic AI inside its own operations and builds it for telecom operators, government and enterprises: agents that read network and business data, open tickets, draft proposals and act through real tools. You will design, build and ship these systems to production.

This is not a research-only role. Your agents will be used every day by our own team and deployed at client sites, often on private infrastructure where data must stay in-country.

At a glance

• Location Dubai, UAE (TECOM)
• Work mode On-site, flexible hours
• Type Full-time
• Experience 3+ years
• Reports to CEO
• Languages English; Arabic a plus

What you will do

• Build LLM agents with tool use, retrieval (RAG) and memory, connected to Odoo, email, network management and monitoring systems
• Design MCP servers and APIs that let agents act safely on real systems
• Build evaluation sets, guardrails and logging so every agent is measurable and auditable
• Deploy on AWS and on-prem NVIDIA GPU for clients who need data sovereignty
• Adapt open models (Llama, Qwen, Mistral) for telecom and Arabic/English use cases
• Work with presales on demos and proofs of concept for operator and government clients

What you bring

• 3+ years software engineering, with at least 1 year shipping LLM applications to production
• Strong Python; TypeScript is a plus
• Hands-on with LLM APIs (Claude, OpenAI or open models), vector databases and agent frameworks
• Docker, Linux, CI/CD and cloud (AWS preferred)
• Clear written English and the habit of documenting what you build
• Inference serving with vLLM or TGI, MLOps
• ERP and Odoo integrations

Your first 90 days

• Ship one production agent that our team uses daily
• Deliver one client proof of concept end to end
• Set up an evaluation harness so we can measure agent quality release by release

What we offer

• Real projects on live operator, government and enterprise networks, not slideware
• Multi-vendor exposure: Viettel, Ericsson, Nokia, Huawei and Open RAN
• AI agents built into daily work, so you spend time on engineering and selling, not admin
• Competitive package based on experience
• UAE employment visa and medical insurance
• Office in TECOM, Dubai, with a small senior team and direct access to the CEO