
Principal AI Architect (w/m/d)
Volkswagen10 days ago
Wolfsburg, Germany +3 moreStaff+
Responsibilities
- Define the technical architecture and reference architectures for CARIAD’s R&D AI solution landscape, including model access and routing, context engineering, retrieval, agent runtimes, orchestration, tool integration, evaluation, observability, APIs, and secure execution.
- Develop reference implementations and prototypes to validate architecture concepts and technical decisions through code, benchmarks, and concrete implementations.
- Establish reusable software components, technical standards, stable interfaces, quality gates, and measurable production-readiness criteria.
- Conduct architecture and code reviews and support engineering teams with cross-team architecture challenges.
- Evaluate build, buy, and reuse approaches based on platform compatibility, integration effort, vendor dependencies, and total operating costs.
- Collaborate with domain engineers and decentralized AI teams to turn business and technical requirements into scalable, production-ready AI solutions.
- Integrate AI capabilities into existing R&D toolchains for requirements engineering, source code, build, test, and validation environments.
- Document architecture decisions, technical trade-offs, and production-readiness criteria, and continuously improve systems based on evaluation results and production insights.
Requirements
- Completed degree in computer science, software engineering, or a comparable field, or equivalent practical experience.
- Demonstrated experience designing and developing complex distributed software systems and owning architecture across multiple teams.
- Practical experience developing and deploying production AI applications, particularly LLM-based, RAG, and agent-based systems.
- Very strong Python programming skills and solid knowledge of API design, cloud-native architectures, distributed systems, and resilient software development.
- Strong understanding of agent-based architectures and agentic software engineering patterns, including tool or function calling, orchestration, state and context management, human-in-the-loop workflows, and controlled execution.
- Experience systematically evaluating LLM and agent-based systems using test datasets, quality metrics, regression testing, LLM-as-a-judge, and quality, latency, and cost benchmarks.
- Experience evaluating and integrating different LLMs, multimodal models, and embedding models, including model-routing and abstraction mechanisms.
- Deep understanding of context engineering, retrieval architectures, embeddings, vector search, reranking, prompt and context management, and structured outputs.
- Strong knowledge of security, privacy, identity and access, guardrails, secure tool usage, and data and model access.
- Experience with AI-specific observability and tracing across models, agents, tool calls, retrieval, and end-to-end execution flows.
- Deep practical experience with at least one modern generative AI or agentic AI technology stack and the ability to evaluate new models, frameworks, and engineering approaches through implementations and benchmarks.
Benefits
- CARIAD emphasizes diversity, individuality, and an inclusive work environment.
- Applicants with disabilities can request application support via careers@cariad.technology.