Senior AI Engineer [RAG Applications (Azure / Kubernetes)]

Context

A large public-sector organization is building a bilingual (French/Dutch) Retrieval-Augmented Generation (RAG) knowledge search service for approximately 10,000 internal users.

This knowledge search platform is the first production workload on the organization's AI Platform and will be followed by additional AI services. The platform combines managed Azure AI services with self-hosted open-source components running on Azure Kubernetes Service (AKS).

The mission is part of a Digital Innovation / AI Center of Excellence (AI CoE) and focuses on the Build phase and early Run phase of the service.

Your Role

As Senior AI Engineer, you will translate the approved High-Level Design into a production-ready, evaluated, and maintainable AI service, working closely with:

• the AI Platform Architect
• the AI Search Solution Architect
• the Digital Innovation / AI CoE team

You are a hands-on senior engineer, responsible for both AI pipeline implementation and cloud-native deployment.

Key Responsibilities

1. RAG Pipeline Implementation (End-to-End)

You design and implement the full RAG flow, including:

• document retrieval and relevance ranking
• LLM orchestration and prompt templates
• citation handling and response grounding
• security trimming and ACL propagation

This includes making concrete design decisions around:

• embeddings
• vector storage
• reranking
• retrieval quality

2. Ingestion & Knowledge Pipeline

You implement and operate the ingestion pipeline, including:

• connectors for Jahia CMS and SharePoint
• document chunking strategies
• embedding generation
• metadata and access control propagation

3. Evaluation & Quality Measurement

You design and implement an evaluation framework for the RAG application, covering:

• definition of quality metrics (retrieval and generation)
• creation or curation of test sets
• automation of evaluation
• integration of evaluation into deployment and iteration cycles

4. Cloud-Native Deployment & Infrastructure as Code

You are personally responsible for deploying the solution using:

• Azure Kubernetes Service (AKS)
• Helm charts or Kubernetes manifests
• Terraform for Azure resources
• Azure DevOps pipelines

You contribute to the AI CoE’s Infrastructure-as-Code standards for application-scoped resources.

5. Early Run & Service Tuning

During the early Run phase, you:

• monitor and tune performance
• improve retrieval and answer quality
• support operational readiness
• contribute to stabilization and continuous improvement

Technology Stack

Cloud & Platform

• Microsoft Azure
• Azure Kubernetes Service (AKS)
• Azure DevOps
• Terraform
• Helm / Kubernetes manifests

AI & Data

• Azure OpenAI Service
• Large Language Models (LLMs)
• Retrieval-Augmented Generation (RAG)
• Vector databases (e.g., Weaviate, Milvus, Pinecone, or equivalent)
• Embeddings & reranking techniques

Development

• Python
• REST APIs
• Git-based workflows

Security & Governance

• Azure API Management
• Azure AI Content Safety
• Security trimming & ACL propagation

Languages : Dutch OR French (Fluent) + English (Fluent)