Forward Deployed

Retrieval Quality & RAG Applications: sections and modules

T6

Retrieval Quality & RAG Applications

Hybrid search, query transformation and reranking, then complete RAG applications with citations, structured data, images and LlamaIndex.

Requirement: Required
For whom: All technical paths
Notebooks: 47
Before this track: RAG Foundations: Ingestion & Indexing
By: Sourav Banerjee

Objectives

Combine keyword and vector search and fuse their rankings
Pick query transformation and reranking from a measured failure
Ship a RAG system that cites its sources per claim
Retrieve over tables, JSON and images as well as text

The model answers from the top-k, so getting the right chunks there decides answer quality, and a product also shows its sources.

T6.1
Retrievers & Hybrid Search
8 lessons
8 lessons
T6.2
Query Transformation & Routing
10 lessons
10 lessons
T6.3
Post-Retrieval Optimisation & Reranking
8 lessons
8 lessons
T6.4
End-to-End RAG Systems
7 lessons
7 lessons
T6.5
Multimodal RAG
5 lessons
5 lessons
T6.6
Framework Fluency: LlamaIndex
6 lessons
6 lessons

Exercise: Take one failing question set and fix it with hybrid search, a query rewrite or a reranker, then add per-claim citations
Done when: You can answer every module's Interview questions without notes