Episode

From Postgres full text search to Retrieval Augmented Generative search

with Adam Wolk

POSETTE: An Event for Postgres 2024

Video of a conference talk about Postgres full text search and retrieval augmented generative search presented by Adam Wolk at POSETTE: An Event for Postgres 2024. This talk iterates through the design of an email search application, starting from basic search through Postgres full text search, semantic search, and finally retrieval augmented generation (RAG), a new AI capability enabled by Azure Database for PostgreSQL. And this talk will introduce you to Milly the Mailiphant, who is ready to help with any Postgres database questions you may have.

Through the session we cover designing the search document, FTS feature inner workings, the pgvector extension to Postgres, plus how to leverage the azure_ai extension to Azure Database for PostgreSQL.

Adam spent 7 years working with Oracle PL/SQL in the financial sector, from which he moved on to using PostgreSQL within a Golf startup for the next 4 years, later became an OpenBSD developer and spent 4 years at a FreeBSD-based company where he mostly worked with C on FreeBSD and began a deep dive into PostgreSQL internals. Long time database & infosec geek and a gamer, currently pushing at the edge of distributed SQL as a Principal Technical Program Manager for PostgreSQL at Microsoft.

Chapters

  • 00:00 - Intro
  • 01:05 - Milly the Mailiphant
  • 01:25 - Basic Search
  • 02:40 - Full Text Search
  • 07:09 - What is Semantic Search
  • 11:56 - Retrieval Augmented Generation (RAG)
  • 13:44 - Demo

Connect

Azure
Azure Database for PostgreSQL