Modular Solutions for Knowledge Management at Scale in RAG Systems

Explore a conference talk from the Toronto Machine Learning Series (TMLS) that delves into building modular document embedding pipelines for efficient knowledge management in RAG (Retrieval-Augmented Generation) systems. Learn how Bell Canada implements customizable processing, ingestion, and indexing solutions to handle diverse document types and use cases in their RAG applications. Discover approaches for managing both batch and incremental updates to knowledge bases, including automatic index updates when documents change in source locations. Understand the integration capabilities with various document sources and the established processes and conventions that ensure effective management and governance of indexes at scale. Through insights shared by Bell Canada's Senior Machine Learning Engineer Adam Kerr and Senior Manager of Artificial Intelligence Lyndon Quadros, gain valuable knowledge about implementing standardized frameworks for enterprise-level RAG applications.