All projects
Shipped·Personal·2023

Train GPT

RAG over unstructured files

/ overview

Translates unstructured files (PDFs, docs, transcripts) into vector embeddings, runs similarity search at query time, and uses top-k context to ground the response.

/ what it does

  • File ingestion pipeline → embeddings
  • Similarity search over Pinecone
  • Context-grounded LLM responses