#EG AI Solution Architect at NCS3

About the role context This role sits embedded inside an AI project — a GenAI initiative within the organisation responsible for delivering business outcomes through AI applications. Each AI project has its own business problems, its own stakeholders, and its own delivery team. Rather than building GenAI capability from scratch, AI projects assemble their solutions on top of common services, building blocks, and reference architecture provided by a central platform team. As the AI Solution Architect for the project, you are the bridge between the business need and a production GenAI solution that uses those shared platform capabilities effectively. The role In this role, you are the AI Solution Architect for one or more AI projects. You translate business and product requirements into actual GenAI solution designs, leveraging the common services, building blocks, and reference architecture provided by the central platform team. You work closely with a platform engagement counterpart — typically a Solution Architect from the central platform team — who helps you apply the platform to your context. You also work hand-in-hand with the project's product, engineering, and delivery teams who turn your designs into production systems. You are a delivery-focused architect. You prefer shipping over standardising; you consume the platform rather than build it; and your success is measured in working solutions for your AI project. Key responsibilities • Own the GenAI solution architecture for your AI project — translating business and product needs into end-to-end designs. • Apply the central platform's reference architecture and consume its common services and building blocks; minimise one-off custom work. • Work hand-in-hand with the platform engagement counterpart to ensure designs leverage the platform effectively, and to surface needs that should become platform features. • Partner with the AI project's product, engineering, and delivery teams to take designs to production. • Make pragmatic trade-offs between time-to-value and architectural quality. • Stay close to the business — understand the use case, the users, and what "good" looks like for your AI project.Technical environment Across all roles, you will be working with — or designing for — the following stack. You do not need every item listed; you should be familiar with most and able to learn the rest. • LLM providers: Commercial LLM APIs and self-hosted open-weight models served via vLLM / LLM-d. • Agent frameworks: LangChain, LangGraph, LlamaIndex. • Retrieval: Milvus and Graph DB, with a custom-built ingestion and retrieval framework on top. • Orchestration and harness: Custom-built harness using Skills, eval loops, prompt and context engineering, and multi-agent orchestration on a service mesh. • Runtime: Containerised applications on OpenShift; multi-cloud Kubernetes (AKS, EKS) deployed via GitOps and ArgoCD. • Guardrails: Custom guardrails plus F5 AI Guardrails, Presidio (PII redaction), and OpenTelemetry plugins. • MLOps: Lightweight, focused on LLM fine-tuning and embedding model lifecycle. (No deep / classical ML.) • Observability: OpenTelemetry, Grafana, Prometheus, Loki, Tempo.How we work • Governance by default: Compliance, security, and guardrails are inherited from the platform, not bolted on per project. • Apps lead, infra supports: GenAI is mostly application-layer work with infrastructure underneath. We expect everyone to be apps-strong and infra-literate. • Telco context: We operate in a telco environment. Telco domain experience is a plus across all roles, not mandatory.