Gen-AI Developer Classroom notes 07/Jul/2026

Embeddings

• Embedding embed the meaning in the form of the vector (a mathematical point)

• Early idea of Embedding was done by a library called as Word2Vec

• Embedding will have the whole vocabulary

• Prompt

IN simple timeline viewer show me evolution of embeddings from word2vec quoting significant milestones

• Embedding models

  • Opensource embedding models
  • Embedding models as service (Cloud)

• Langchain Embeddings the base class Embedding which is in langchain-core.embeddings which has two simple methods

  • embed_query
  • embed_documents

Vector Database (Vector Store)

• Once we have vectors we need to store them Refer Here to understand vector databases
• Prompt

Compare with Relational Database and Explain the organization and operations possible on vector databases

• Vector Database Options

  • Developer:
    • Chroma
    • FIASS
  • PROD:
    • Familiar:
      • pgvector
      • mongodb
    • Cloud providers:
      • AWS
      • Azure
      • GCP
    • SaaS Providers:
      • Pinecone
    • Opensource Vector database:
      • Weaviate

• Langchain vector databases integration Refer Here

• Exercise -> Indexing Pipeline:

  • Create a folder with a text files
  • Each text file contains information about some city
  • Ensure text is atleast 200 lines organized as paragraphs
  • Use RecursiveCharacterSplitter
  • Load and split => documents
  • Use text-embedding-005 and chroma to store in vector database in some folder.
  • Ensure every chunk has the following metdata
    • city: <>
    • chunk-id:
  • Vector database => ./data/example1