Conversation AI Engineer

Atribs Metscon Group, an Abu Dhabi–based enterprise AI and technology consultancy, is looking for a Conversation AI Engineer to join onsite in Abu Dhabi. The company serves regulated enterprise clients across banking, energy, healthcare and government, with projects that span the GCC, the United Kingdom and Germany. They deliver implementation-led engagements and work in environments from on-premises to private and public clouds, aiming to leave clients with systems they can run themselves. If you enjoy turning complex requirements into reliable, production-ready AI that people can use, this role could suit you.

The role centers on building customer-facing AI agents that operate in English and Arabic and are accountable for their behavior in production. The first deployment for this position is a bilingual customer service agent for a major UAE energy client, covering web, messaging and voice channels. You’ll work closely with client teams in a hands-on, on-site setting to ensure a solid, operable solution from day one.

The team values deployment flexibility and seeks engineers who can adapt to sovereign air-gapped, on-premise, private cloud or public cloud environments. This is a hands-on, client-facing opportunity for someone who wants to craft real-world AI systems that deliver measurable outcomes and empower the client to operate and extend what’s built.

Making bilingual conversations count in the real world

In this role you’ll design and build AI agents that resolve customer inquiries using approved knowledge sources, with every answer traceable to its source content. You’ll own the quality of retrieval, how content is prepared, chunked, indexed and refreshed, and you’ll define how the agent should behave if it encounters questions it can’t answer well.

Part of the job is shaping bilingual behavior so the system can detect languages, switch mid-conversation when needed, and produce Arabic responses that feel natural to a Gulf audience. You’ll also set clear escalation rules that hand customers off to human agents when appropriate, ensuring no one gets stuck in a loop.

You’ll establish evaluation plans before launch, including test sets, accuracy measurements and regression checks for every change. Instrumentation will surface metrics such as resolution rate, escalation volume and response accuracy so the client can see progress in real time. Work will take place in the client’s environment under their security and data residency requirements, and you’ll hand over documentation plus training so the client’s team can operate and extend the agent afterward.

What the job requires, and what’s nice to have

• Three or more years building software, with at least a year working on language model based applications used by real users
• Hands-on production experience with retrieval over enterprise content and a clear view of why retrieval quality can fail
• Strong Python skills and comfort coding inside someone else’s codebase and cloud account
• Experience designing conversation flows that include failure paths, not just the ideal path
• Evidence of actually measuring an AI system rather than merely demonstrating it
• Ability to work onsite with client teams and explain technical decisions to non technical stakeholders

• Arabic language skills, especially Gulf dialect handling in speech or text
• Experience with contact centre platforms, WhatsApp Business or voice channels
• Work in a regulated environment such as banking, energy, healthcare or government
• Exposure to guardrails, prompt injection defence and content safety in customer facing systems

Tips to stand out for this Conversation AI Engineer role

Lead your resume with your software and AI work that actually shipped: highlight projects where you built LLM-based applications used by real users, especially those with bilingual capabilities and production-grade retrieval from enterprise content. Show the sensors and metrics you put in place to prove success, not just the code you wrote.

When you describe your must-have skills, give concrete examples. Put a project front and centre that demonstrates a bilingual agent with language detection, mid-conversation switching and Arabic responses tailored for a Gulf audience, plus how you measured its effectiveness in a live environment.

For interviews, be ready to explain decisions to non-technical stakeholders and to walk through collaboration with client teams onsite. Talk about how you set up evaluation before launch, what test sets looked like and how you tracked accuracy and regression across changes.

When you ask questions, target the client’s 90 day plan: how the team defines a working agent in the client environment, what the agreed accuracy bar is, and how escalation rules will be signed off by the client’s service team. You can also ask about the client’s security and data residency requirements and how on-site work is scheduled.