Build and maintain LLM‑powered features end‑to‑end (prompting, RAG pipelines, structured extraction/classification such as entity extraction).
Develop data ingestion, cleaning and indexing pipelines for RAG, including n8n workflows for intake and enrichment (connectors, transformations, error handling, scheduling).
Contribute to lightweight model tuning and systematic evaluation.
Establish evaluation and observability for RAG (dashboards, automated reporting, experiment tracking) to ensure reliability and factual grounding.
Optimise prompts, retrieval and context strategies to improve accuracy, reduce hallucinations and control latency/cost.
Work hand‑in‑hand with our ML/DevOps engineer to ensure smooth deployments, reliability and continuous improvement.
Coordinate and provide technical guidance to a small offshore AI development team (clear specifications, code reviews, quality standards), with support from the Head of AI.
Collaborate with product and engineering to scope and deliver incremental value in short, iterative releases.

Strong Python skills; experience with PyTorch or Transformers; familiarity with the Hugging Face ecosystem.
Practical knowledge of LLM tooling (e.g. LangChain or LangGraph) and RAG concepts.
Experience building on AWS, ideally including:
Serverless functions (AWS Lambda) for orchestration,
Elastic compute (EC2) for workloads,
Foundation model services (Bedrock or SageMaker) for model hosting and tuning.
Hands‑on with n8n; workflow automation experience is a plus.
Containerisation with Docker; proficiency with Git and CI/CD.
MLOps fundamentals: MLFlow for experiment/model tracking; evaluation frameworks (e.g. RAGAS).
Clear communication, collaborative mindset and focus on shipping.
Languages: strong English; German is a plus. Applications in English or German are welcome.
Education: Degree in Computer Science or equivalent preferred; equivalent practical experience acceptable.

AI/ML Engineer, LLMs, RAG

Key skills