AI Engineer & Agentic Developer

On behalf of our client, an international industrial group, we are looking for an AI Engineer to join its central Data Science & AI Engineering team. The team is part of corporate R&D and delivers projects for business units worldwide. It combines specialists in agentic AI, time series and computer vision, and works directly with scientists, process engineers and business owners. The team runs its own agentic platform on Azure with a custom agent harness.

It already supports several use cases in production, from finance processes to research assistants that query scientific and operational databases. The next phase is underway: more use cases in parallel, more data sources connected, and a larger library of shared skills and tools. You will build those agents and strengthen the platform underneath them.

What you will do: In agent development, you will design and build agents that carry complex, domain-specific logic, translating expert knowledge from scientists and engineers into reasoning steps, tools and guardrails; build multi-step and multi-agent workflows with LangGraph, combining retrieval, tool calls and business rules; and connect agents to structured enterprise data in Databricks, often combining several sources in one solution. In context and knowledge management, you will define what context an agent needs to reason correctly: instructions, retrieved documents, table and column descriptions, domain rules; improve metadata on the data platform so tables and fields are described well enough for agents to use them reliably; and design retrieval and grounding strategies that keep answers traceable to their source.

In skills and tooling, you will build reusable skills and tools that agents across use cases can share, decide which agent gets which tools and permissions, and keep the tool library maintainable. In evaluation and quality, you will set up structured evaluation of agent behaviour: test sets, metrics, regression checks and human review; use evaluation results to improve prompts, tools and architecture, and report quality clearly to stakeholders. In delivery and collaboration, you will deliver solutions on Azure and Databricks, with IT owning integration and operations; contribute to platform architecture decisions and engineering standards; and follow developments in LLMs and agent frameworks and assess what is worth adopting.

Requirements: Master's degree in Computer Science, or a PhD in Mathematics or a related quantitative field; 3+ years of hands-on software or AI engineering, including building agentic or LLM-based systems; at least one agentic application delivered to production that goes well beyond a chatbot or summarisation tool; real production experience on Azure and Databricks; strong Python and solid software engineering practice (testing, version control, code review, clean architecture); hands-on experience with LangGraph or a comparable orchestration framework, and with OpenAI or similar models; experience with RAG, tool calling and structured evaluation of LLM output; backend and data focus, comfortable working with data platforms, APIs and system architecture; and clear communication in English with technical and non-technical stakeholders.

Nice to have: experience in an industrial, manufacturing, chemical or process environment; Azure OpenAI, MLflow, Unity Catalog; experience building shared platforms or internal developer tooling.

What you get: complex, company-specific problems with access to real industrial data; room to explore and build, without operational load (once a solution is built, IT takes over operations); a multidisciplinary team where you learn from colleagues in vision AI, time series and process modelling; and involvement in research projects with external partners.

Offer: Location Olen, hybrid (3 days per week on site), language English, experience 3+ years, start as soon as possible, 6 month contract.

Note: the structured metadata indicates a minimum of 5 years of experience, while the description above states 3+ years.