Senior AI Platform Engineer - LLMs, GPUs & Kubernetes

IC; Dubai; On-site; AED 420,000 - 620,000; Full time

Job summary

IC Markets Global in Dubai is seeking a Senior LLMOps / AI Platform Engineer to build, deploy, and operate production-grade LLM infrastructure. You will focus on latency-optimized inference, GPU memory management, and scalable AI services on Kubernetes and AWS.

The role covers observability, RAG, embeddings, vector databases, and CI/CD pipelines, with hands-on work on LangChain/LangGraph tooling and self-hosted models like vLLM and Hugging Face.

Qualifications

• Strong Python and FastAPI experience.
• vLLM, Hugging Face and self-hosted LLM deployment.
• Kubernetes, Docker, Helm and AWS experience.
• NVIDIA GPU inference and performance optimization would be a plus.

Responsibilities

• Deploy and operate self-hosted LLMs using vLLM, SGLang, Ollama.
• Optimize LLM inference for latency, throughput, concurrency, GPU memory, KV cache, and cost.
• Manage GPU workloads across multiple NVIDIA GPUs.
• Deploy and maintain AI services on Kubernetes / AWS EKS using Docker and Helm.
• Implement LLM reliability mechanisms including health checks, monitoring, automated recovery, and model restart/refresh strategies.
• Implement observability using Langfuse/LangSmith, OpenTelemetry, Prometheus, and Grafana.
• Deploy and optimize RAG systems, embedding models, and vector databases such as Qdrant, Milvus.
• Support AI agents and workflows built with LangChain and LangGraph.
• Build and maintain CI/CD pipelines for AI services and infrastructure.
• Troubleshoot production issues across LLMs, GPUs, Kubernetes, networking, and AI applications.

Skills

Python

FastAPI

LLM/AI Platform eng

LangChain / LangGraph / LangSmith

Production troubleshooting

Tools

Kubernetes

Docker

Helm

AWS

vLLM / Hugging Face self-hosted

Qdrant / Milvus

PostgreSQL

Redis

GitHub Actions / CI/CD

Job description

IC Markets Global in Dubai is seeking a Senior LLMOps / AI Platform Engineer to build, deploy, and operate production-grade LLM infrastructure. You will focus on latency-optimized inference, GPU memory management, and scalable AI services on Kubernetes and AWS.

The role covers observability, RAG, embeddings, vector databases, and CI/CD pipelines, with hands-on work on LangChain/LangGraph tooling and self-hosted models like vLLM and Hugging Face.