AI Architect (LLM)

About the role

We are looking for an AI Architect specialized in Large Language Models (LLM) to design, develop and productionize LLM‑based and Retrieval‑Augmented Generation (RAG) solutions for an aviation‑industry client.

Key responsibilities

• Design, develop, test and productionize LLM‑based and RAG solutions for enterprise use cases.
• Lead the development of internal chatbots, conversational AI, knowledge assistants and agentic AI products from proof‑of‑concept to production.
• Own the technical architecture of enterprise LLM solutions, including model selection, deployment, serving, routing, evaluation and monitoring.
• Deploy, operate and optimise open‑weight and commercial LLMs on on‑premise or private infrastructure, including setting up Red Hat OpenShift AI platforms.
• Evaluate and select models based on quality, latency, throughput, cost, security and licensing constraints.
• Optimise LLM inference through quantisation, batching, caching, GPU resource management and workload allocation.
• Design model and semantic routing mechanisms and support agentic and tool‑calling architectures.
• Collaborate with platform, data, software engineering and business teams to translate requirements into scalable AI solutions.

Required profile

• Minimum 7 years of professional experience in AI, ML, data science, software or AI platform engineering.
• Strong hands‑on experience taking LLM systems from experimentation to production in high‑traffic, high‑concurrency environments.
• Deep understanding of GPU‑based LLM inference, memory utilisation, throughput and latency optimisation.
• Proven experience with containerised orchestration (Kubernetes/OpenShift) and on‑premise deployment.
• Experience designing and improving RAG‑based applications, embeddings, vector search and retrieval quality optimisation.
• Ability to provide technical leadership, architectural guidance and mentor engineering teams.

Required skills

• Python programming
• Kubernetes and OpenShift orchestration
• Red Hat OpenShift AI (preferred)
• Hugging Face, vLLM, NVIDIA inference, TGI, Triton
• GPU‑based LLM inference optimisation (quantisation, batching, caching)
• Model serving and routing, semantic routing, multi‑model architectures
• Retrieval‑Augmented Generation (RAG), embeddings, vector databases
• Agentic AI workflows, tool/function calling
• CI/CD, version control, API design and testing

What we offer

• Open communication, flexibility and start‑up spirit
• Learning & development opportunities and company‑paid professional certificates
• Access to online training platforms (Udemy, Pluralsight, Coursera, etc.)
• Dynamic work ecosystem with initiative and responsibility
• International project exposure
• Private health insurance and birthday leave policy