Validation Engineer - Context, Semantics & Memory (f/m/x)

Our platform teams build AI-based toolchains that simulate, reprocess and train models on petabyte-scale automotive data - and we increasingly rely on agentic AI to operate, maintain and improve these toolchains. The glue that makes agents effective is context: the right knowledge, delivered at the right time, in the right shape. We are looking for an engineer who can design and evolve the knowledge-management and context-engineering backbone that powers both our product (an inter-domain multi-agent vehicle system) and our internal development toolchain.

What awaits you?

• You design and operate knowledge graphs and graph databases that capture domain knowledge, toolchain topology, operational incidents, root-cause analyses and lessons learned - enabling graph-based retrieval across the organization.
• Further, you build context-engineering pipelines that dynamically assemble, rank and budget the information supplied to LLM-based agents.
• Also, you implement feedback loops: coding agents and skills that interact with our toolchain log outcomes, diagnose failures and write findings back into the knowledge graph, making the system progressively more capable.
• It is your responsibility to develop and maintain ontologies and semantic models for the automotive AI domain - sensor configurations, simulation scenarios, data-pipeline topology, model lineage - ensuring consistent vocabulary across teams and tools.
• You architect agent memory systems (working, episodic, semantic) with appropriate consolidation and retention mechanisms so that long-running agent sessions retain useful experience without context-window bloat.
• In addition, you evaluate and optimise retrieval quality using relevance, coverage, freshness and redundancy metrics, and run experiments comparing retrieval strategies (dense, sparse, graph, hybrid).
• You contribute to cost-efficient model selection and context-compression strategies (prompt compression, semantic caching, model routing) to keep latency low and budgets sustainable at scale.

What should you bring along?

• University degree in Computer Science, Information Science, AI/ML, or a related field.
• 1-3 years of professional experience in knowledge management, knowledge-graph engineering or context engineering for AI/LLM systems.
• Expert-level programming skills in at least one high-level language (C++, Java, Kotlin or Python).
• Proficiency in AI-assisted development using coding agents such as Claude Code, GitHub Copilot or OpenCode.
• Hands-on experience building and querying knowledge graphs (RDF/SPARQL or property graphs) and integrating them with LLM-based retrieval (e.g. GraphRAG, hybrid search).
• Experience with systems design, particularly for knowledge-intensive or retrieval-heavy architectures.
• Familiarity with agent memory architectures, ontology design, semantic modelling, graph-based analytics, entity resolution or knowledge-graph embeddings.

You enjoy working in an international team and are passionate about software quality? Apply now!

Find out more about Artificial Intelligence at the BMW Group here.

If you apply, the next stages of the recruiting process include an online test followed by technical interviews with the hiring manager (either virtual or in person).

You can find helpful tips on your application and the application process here.

At the BMW Group, we place great importance on equal treatment and equal opportunities. Our recruiting decisions are based on the personality, experience, and skills of the applicants. Learn more here.