Required skills
- Software Development
- Python
- RAG
- LangChain
- LlamaIndex
- Vector Database
- Vector Search
- Prompt Engineering
- Quality Assurance
- Communication
Languages: German
Job description
We are looking for experienced freelancers in RAG/Prompt Engineering to design and implement RAG pipelines for customer projects, focusing on data preparation, prompt writing, and quality testing.
Project overview
A language model is only as good as the knowledge it receives at the right moment. Whether an AI assistant correctly states a delivery time or makes something up is decided during chunking, searching, and in the prompt. This is exactly where we are looking for reinforcement as experienced freelancers who not only know Retrieval-Augmented Generation from tutorials but have already used it productively. The collaboration is 100% remote, from Germany, Austria, or Switzerland, starting November 11, 2026.
Responsibilities
You will conceptualize RAG pipelines for customer projects: connect data sources, prepare documents, and divide them into meaningful sections, select embeddings and search. You will implement the pipelines with LangChain, LangGraph, or LlamaIndex, depending on what fits the project. You will write and version system and user prompts, ensuring that answers remain verifiable and the model openly admits knowledge gaps. You will build test sets with typical customer questions and measure hit quality and answer quality before anything goes live and after every change.
You will weigh with us whether a cloud model like Claude, GPT, or Mistral is suitable or if a local model via Ollama is more sensible.
Requirements
Several years of experience in software development, preferably with Python, and at least one RAG system that you have brought into operation. Confident handling of LangChain or LlamaIndex as well as at least one vector database or vector search. A methodical understanding of how to measure the quality of answers instead of relying on gut feeling. The ability to explain technical trade-offs in a way that customers without a technical background can understand. Confident communication in German or English. Own self-employment or freelancer status based in Germany, Austria, or Switzerland.
Nice to have, but not a must: Experience with LangGraph or agent frameworks like CrewAI, with Hugging Face, or with operating local models.
Project details
- Start: flexible
- Location: DE
- Languages: German
Indexed by 9am's meta-search. You are a job provider? Claim or manage this listing
Found on interlance.de