AI Engineer

Il y a 23 heures

Arrondissement of BrusselsCapital, Brussels, Belgique Adjugo Temps plein
pbDates: /b
- /p pbArrangements: /b Full time; hybrid; on-site in Brussels 2 to 3 days per week. /p pSTIB-MIVB is building employee
- and customer-facing AI services across the organization. Services range from retrieval-based assistants to AI agents that call tools and business systems on users' behalf. They run on the STIB AI Platform. The first services go live from December 2026, with more following in 2027. /p pThe Medior AI Engineer joins the AI Center of Excellence's Digital Innovation AI delivery team and owns production operations. Each service is handed over by the Senior AI Engineers who built it. The engineer keeps services reliable, measured, secure and cost-efficient. Collaboration includes the internal AI Platform Architect, two Senior AI Engineers responsible for Build, business and content owners, and IT operations teams. /p pThe platform combines managed Azure services with self-hosted open-source components on AKS. Microsoft Foundry uses the OpenAI model family for LLMs and embeddings, alongside Azure AI Search, Azure API Management and Azure AI Content Safety. Python and LangChain support agent orchestration and an ingestion pipeline, with MLflow on Databricks for evaluation and OpenTelemetry tracing. The team owns application-scoped Azure Infrastructure-as-Code through Terraform and Azure DevOps pipelines; small enhancements belong to Run, while structural changes and new features stay with Build. /p h3Top Reasons to Apply /h3 h3Production AI ownership /h3 h3Agents and RAG /h3 h3Public transport impact /h3 h3Cross-functional AI collaboration /h3 h3Responsibilities /h3 poperate the AI services in production: monitor availability, latency and errors, handle incidents and service requests, analyse root causes, then fix or elevate ; /p pkeep answer quality under control: run and maintain evaluation sets (golden sets) with business content owners, track answer quality, non-answer rate and user feedback, and detect and investigate quality regressions ; /p poperate the ingestion pipelines: incremental refresh, failed documents, index health, and onboarding of new content sources or business domains within existing connectors ; /p poperate agent integrations: monitor tool and integration calls to business systems, handle failures and timeouts, and review action logs and exceptions for agents that act on users' behalf ; /p pmanage change safely: model version upgrades and retirements, prompt and configuration changes, dependency and security updates, each validated against the evaluation sets before release ; /p pfollow run cost: token consumption and cost per service, alerting on drift and proposing optimisations ; /p pmaintain the Terraform, Helm and Azure DevOps artefacts, and keep the DEV, TST, ACC and PROD environments consistent ; /p papply data protection rules: trace retention, erasure and access restrictions, as defined in each service's data protection impact assessment ; /p h3Must Have /h3 pRAG (Retrieval-Augmented Generation) intermediate proficiency /p h3Detailed Responsibilities and Skills /h3 h3Additional Responsibilities /h3 ullimaintain runbooks and documentation, and report service health to service owners and the AI CoE on a regular basis ; /lilideliver small enhancements and fixes; structural changes and new features stay with the Build team. /liliMedior Artificial Intelligence (AI) Engineer; most recent

experience:
this year (2026). /liliDutch or French: native or bilingual proficiency (one of the two languages). /liliEnglish: full professional proficiency. /liliOn-site presence in Brussels 2 to 3 days per week. /li /ul h3Technical skills and tools /h3 ulliAzure DevOps: intermediate level; most recent

experience:
this year (2026). /liliAzure Kubernetes Service
- AKS: intermediate level; most recent

experience:
this year (2026). /li liAzure OpenAI Service: intermediate level; most recent

experience:
this year (2026). /li liLangChain: intermediate level; most recent

experience:
any time. /li liLLM / Large Language Models: intermediate level; most recent

experience:
this year (2026). /li liPython: intermediate level; most recent

experience:
this year (2026). /li liRAG (Retrieval-Augmented Generation): intermediate level; most recent

experience:
this year (2026). /li liTerraform: intermediate level; most recent

experience:
this year (2026). /li liVector databases: intermediate level; most recent

experience:
this year (2026). /li /ul h3Selection questions — status not stated as mandatory /h3 ulliQuestion 1 screens for at least 6 months personally operating or supporting a production LLM application (RAG, agent or document processing). /li liQuestion 2 asks for personal investigation of an LLM or agent quality issue or incident using traces or logs. /li liQuestion 3 asks for running or maintaining LLM evaluations, including test sets after model or prompt changes. /li liQuestion 4 asks for personal deployment of Kubernetes manifests or Helm charts and Terraform code through Az