RAG & Knowledge Systems
Priya Raman
Priya teaches retrieval-augmented generation from the data up: parsing, chunking, retrieval and the evaluation that tells you whether any of it is working.
Sample profileLoading page
RAG
RAG & Knowledge Systems
Build retrieval pipelines that return grounded, cited answers.
- Level
- Intermediate
- Format
- Online · Mentor-guided
- Duration
- 8 weeks (sample schedule)
- Effort
- 5–7 hrs/week (suggested)
- Modules
- 5 modules
- Price
- Price on request (demo)
- Availability
- Open enrollment (demo)
Register interestEditable sample values
01
Overview
Most useful LLM applications need to answer questions about information the model was never trained on. This program teaches you to build retrieval systems that find the right passages, pass them to the model, and return answers your users can verify.
You will work through each stage of a RAG pipeline as an engineering problem: parsing messy source documents, choosing chunk sizes, comparing embedding models, combining keyword and vector search, and reranking candidates. Every change is measured against a retrieval test set, so you can see what actually improved.
Mentor reviews focus on your design decisions — why you chose a chunking strategy, how you handle documents that change, and what your system does when the answer is not in the corpus.
fig. — retrieval-augmented generation
Ingestion · offline
documents/
PDF, Markdown, HTML
parse + clean
keep headings & ids
chunk
structure-aware, overlap
embed
vector per chunk
index
vectors + metadata
shared index
Query · online
question
user input
retrieve top-k
hybrid: vector + keyword
rerank
keep the best 4
generate
answer only from sources
answer + [1][2]
citations to chunks
02
Prerequisites
- Comfortable writing Python and working with third-party packages
- Basic SQL and familiarity with a relational database such as PostgreSQL
- Experience calling a model API, or completion of LLM Foundations for Developers
- Understanding of JSON and REST APIs
03
Curriculum
Module 01: Ingestion and Document Processing
Turn raw source material into clean, structured text with the metadata retrieval depends on.
- 01.1Parsing PDFs, HTML, and Markdown without losing structure
- 01.2Preserving metadata: titles, sections, dates, and access permissions
- 01.3Incremental ingestion and handling updated or deleted documents
- 01.4Deduplication and basic content-quality filters
Module 02: Chunking and Embeddings
Split documents into retrievable units and represent them as vectors you can search.
- 02.1Fixed-size, recursive, and structure-aware chunking
- 02.2Chunk size, overlap, and their effect on recall
- 02.3How embedding models work and how to compare them on your data
- 02.4Storing and indexing embeddings with pgvector
Module 03: Retrieval and Ranking
Find the most relevant passages with a combination of search techniques.
- 03.1Vector similarity search and approximate nearest-neighbor indexes
- 03.2Keyword search with BM25 and why it still matters
- 03.3Hybrid search: combining scores with reciprocal rank fusion
- 03.4Reranking candidates with a cross-encoder
- 03.5Metadata filtering and query rewriting
Module 04: Grounded Generation and Citations
Assemble context, generate answers, and point every claim back to its source.
- 04.1Context assembly within a fixed token budget
- 04.2Prompting for grounded answers and explicit refusals
- 04.3Returning structured citations to source passages
- 04.4Handling questions the corpus cannot answer
Module 05: Evaluating Retrieval Systems
Measure retrieval and answer quality separately so you know which stage to fix.
- 05.1Building a question–passage test set
- 05.2Retrieval metrics: recall@k, MRR, and hit rate
- 05.3Groundedness and citation-accuracy checks
- 05.4Running retrieval regression tests in CI
04
Skills and project outcomes
Skills you’ll practice
- Document parsing and ingestion pipelines
- Chunking strategy design
- Embedding model selection and comparison
- Vector search with pgvector
- Hybrid search and reranking
- Citation-backed answer generation
- Retrieval evaluation with recall@k and MRR
What you’ll build
- A document ingestion pipeline with incremental updates
- A hybrid search service with reranking, exposed through an API
- A knowledge assistant that returns answers with source citations
- A retrieval test set and evaluation report comparing at least two configurations
Related project briefs
Outcomes describe skills and artifacts you will practice. No certification, employment or career outcome is guaranteed.
05
Mentors
Evaluation & AI Reliability
Jordan Lee
Jordan teaches how to know whether an LLM feature is good enough to ship — and how to notice when it stops being good enough.
Sample profile06
Program FAQs
Not for this program. Exercises use PostgreSQL with pgvector, which is enough for most small and medium corpora. We discuss when a dedicated vector store becomes worth the extra operational overhead.
07
Where this fits
Interested in RAG Systems?
Register your interest and tell us about your experience. We’ll suggest where to start.