Training project

RAG assistant over document corpus

Related course: Generative AI

Problem

Product teams want assistants grounded in private docs, not unconstrained LLM answers—retrieval quality matters.

Architecture (high level)

Document loaders/splitters → embeddings → vector store (e.g. Pinecone/Chroma) → retrieval → LLM generation via LangChain; evaluation and cost-aware API design from RAG and Production modules.

Technologies

  • LangChain
  • OpenAI API
  • Pinecone
  • Python
  • FastAPI

This is a Vector Skill Academy training project completed during the course. It is not confidential client work and does not represent a live production engagement.

More projects in this course