I will design your ai rag knowledge base and vector database backend

B
bistcon
B
bistcon
Paul S

About this gig

A RAG system is only as good as the knowledge behind it. Most fail on the data layer wrong chunking, no metadata filtering, weak retrieval.


I'm Paul-Ferdinand Steuck: 7 years IT consulting, MSc Information Systems (1.1), PhD researcher on generative AI and LLMs.


I design and build the knowledge backbone: which data exists, how it fits together, what's actually suitable plus metadata filtering, lexical + semantic search, reranking and hosting. I work with vector databases like Qdrant and Pinecone, Python/Rust ETL pipelines and structured methods (requirements analysis, BPMN, UML). I also advise on the real trade-off: RAG vs. fine-tuning vs. other approaches.


All work under NDA, fully confidential & encrypted.


Please contact me before ordering so we can scope your data and goals.

Get to know Paul S

Paul S

IT Consulting and PhD Candidate

  • FromGermany
  • Member sinceDec 2020
  • Avg. response time1 hour
  • Languages

    English, German, French
I help companies build chatbots, RAG systems and multi-agent AI that work — by getting the concept right before anyone writes code. 7 years in IT consulting (Germany), MSc Information Systems (1.1), PhD researcher on generative AI, LLMs and conversational agents. Plenty of people can code an LLM app. The hard part is requirements analysis, the right tools and clean data — that's my focus. I use BPMN, UML and a current stack: LangChain/LangGraph, OpenAI Agents SDK, Pydantic AI, Qdrant, Pinecone, Python/Rust ETL, APIs. All projects under NDA. Please reach out before ordering.