IRL & Online Program
Advanced RAG & Enterprise Data
Move beyond basic vector search. Master GraphRAG, hybrid retrieval, and chunking strategies for enterprise knowledge bases.
Hands-on Highlights
- Implement Hybrid Search (Dense + Sparse retrieval)
- Build a Knowledge Graph using GraphRAG
- Improve accuracy with Cross-Encoder reranking
- Evaluate RAG pipelines using RAGAS
Detailed Syllabus
Week 1-2
Advanced Chunking & Indexing
- Semantic vs structural chunking
- Hierarchical indexing (Parent-Child document retrieval)
- Handling PDFs, tables, and messy enterprise data
- Updating and deleting vectors efficiently
Week 3-4
Hybrid Search & Reranking
- BM25 keyword search vs Vector search
- Combining scores with Reciprocal Rank Fusion (RRF)
- Using Cohere or BGE Cross-Encoders for reranking
- Optimizing retrieval latency
Week 5-6
GraphRAG & Structured Data
- Introduction to Knowledge Graphs (Neo4j)
- Extracting entities and relationships with LLMs
- Combining Graph traversal with vector search
- Text-to-SQL for structured database querying
Week 7-8
Evaluation & Guardrails
- Evaluating RAG with RAGAS (Faithfulness, Relevance)
- Implementing NeMo Guardrails for safety
- Preventing data leakage and securing prompts
- Capstone: Enterprise Knowledge Search Engine
Target Roles & Career Paths
Search & Relevance Engineer
Enterprise AI Architect
Data Engineer
These are the primary roles you will be equipped to apply for upon successful completion of the course and portfolio projects.