Job description Context A global leader in materials technology and recycling, headquartered in Belgium, focuses on clean mobility and circular economy solutions by developing high-performance materials for rechargeable batteries, automotive catalysts, and precious metals recycling. With a strong commitment to sustainability, the company aims to reduce harmful emissions, power the vehicles of the future, and give new life to used metals. The Computational Science - Data Science team (part of the Corporate Research & Development department) located in Belgium & Germany is expanding its agentic AI capability to support a growing internal customer base. The team has an established framework and supporting expertise (data scientists, tooling, and ways of working) and is now looking for hands-on builders who can translate real business needs into working AI agents. The assignment is based in Hanau, Germany, with strong local stakeholder interaction and a practical delivery focus (prototype, validation, productionisation). Missions Translate business context (process steps, decision rules, documents, and data) into structured agent designs: tools/functions, workflows, guardrails, and evaluation criteria. Build, iterate, and harden AI agents using the existing framework and tooling (including agent-to-agent collaboration patterns where required). Implement retrieval and context strategies so agents can reliably use internal knowledge sources (documents/data) while maintaining quality and traceability. Integrate agents with relevant systems and interfaces (APIs, data sources, process triggers) to enable autonomous task execution where feasible. Define and execute testing approaches (functional, robustness, regression) and support user validation with business owners. Support production-readiness activities: monitoring/logging concepts, error handling, prompt/version control, deployment collaboration (DevOps where applicable). Create clear documentation and knowledge transfer materials so solutions can be maintained and scaled across more internal customers. Collaborate with the team and stakeholders to drive adoption and ensure solutions fit operational reality. Tools & Environment Agentic AI frameworks and tooling, including LangGraph (preferred) and/or LangChain for agent/workflow orchestration. GPT-family / LLM APIs or comparable models. Python programming language with clean code practices, modular design, and version control (Git). Integration with data sources and enterprise artefacts (documents, structured data, process data) via lightweight connectors/APIs. Reliability patterns for LLM applications: prompt/versioning, structured outputs, fallbacks, caching, observability, evaluation. DevOps/MLOps exposure (CI/CD, containerisation, environment management, deployment pipelines, monitoring) is a plus. Agile delivery setup (Scrum/Kanban) and iterative product development methodologies. Working Conditions Location: Hanau, Germany On-site presence required: 3 days per week following client’s guidelines Contract type: Temporary Start date: ASAP (targeting end of May / beginning of June) Duration: Until end of the year, extension possible based on results Compensation currency: EUR Additional Information Language proficiency: English fluent (spoken and written); German and other languages are nice to have. Safety equipment: Not applicable (note that for other sites a neon vest is mandatory). Travel: None required (0%). NDA: Not applicable. SCC (VCA) certificate: Not applicable. Requirements Strong hands-on experience building agentic AI solutions (multi-step reasoning workflows, tool/function calling, orchestration, evaluation) Practical experience with LangGraph (preferred) and/or LangChain for agent/workflow orchestration Experience working with GPT-family / LLM APIs or comparable models, including context design, grounding strategies, and output control Solid software engineering fundamentals : Python (preferred), clean code practices, modular design, version control (Git) Experience integrating agents with data sources and enterprise artefacts (documents, structured data, process data) and building lightweight connectors/APIs Understanding of reliability patterns for LLM apps: prompt/versioning, structured outputs, fallbacks, caching, observability, evaluation Nice to have: DevOps/MLOps exposure (CI/CD, containerisation, environment management, deployment pipelines, monitoring) Education: Bachelor’s/Master’s in Computer Science (preferred) or broader IT/engineering background; equivalent experience accepted Strong stakeholder management : able to sit with business users, ask the right questions, and translate ambiguity into actionable build steps Workshop facilitation mindset with pragmatic focus on outcomes Ability to work on-site, hands-on, and iteratively with domain experts; comfortable in operational environments Ownership and delivery drive : proactive, structured, and able to move from prototype to production-quality implementation Clear communicator who can explain limitations/risks and propose workable alternatives without overpromising Experience working in an Agile delivery set-up (Scrum/Kanban) and iterative product development Nice to have: certifications or formal training in cloud/AI engineering (e.g., Azure/AWS), DevOps, or software architecture English: fluent (spoken and written) German: nice to have