RAG Architectures — Foundations
A 60-minute engineer’s tour of retrieval-augmented generation — embeddings, vector stores, hybrid search, re-ranking, and the failure modes nobody puts in the slides.
Last updated: 2026-05-19
What you'll learn
By the end of this course you'll be able to:
- When RAG is the right pattern — and the three cases where it isn’t
- How embeddings actually work, and why “more tokens” isn’t always better
- Vector stores compared — pgvector, Pinecone, Azure AI Search, Weaviate
- Hybrid search done right: BM25 + vectors + semantic ranking
- Re-ranking strategies (cross-encoders, LLM rerankers, hybrid) and their cost curves
- Evaluation — retrieval quality vs answer quality, and how to measure both
Who this is for
Backend engineers, ML engineers, and applied scientists building retrieval-augmented systems. Especially valuable for engineers shipping their first production RAG system — or inheriting a flaky one and trying to figure out why retrieval quality looks great on the slide deck and terrible on real user queries.
Curriculum
8 chapters · 2 hands-on exercises · capstone challenge
Each chapter ends with the learning objectives ticked off. Quizzes are auto-graded with feedback; exercises are open-ended and produce artifacts you can take to your team.
1. Why RAG — and when not to use it
- Distinguish RAG from long-context, fine-tuning, and tool-use approaches
- Recognize the 3 use cases where RAG is the wrong architecture
2. Embeddings explained
- Build a working mental model of dense vector representations
- Pick embedding models with the right context length, dimension, and language coverage
3. Vector stores
- Compare pgvector, Pinecone, Azure AI Search, and Weaviate on latency, cost, and ops
- Choose the right store for your scale and operational footprint
4. Hybrid search
- Combine BM25 keyword search with vector search and semantic ranking
- Tune fusion strategies (RRF, weighted) for your query distribution
5. Re-ranking strategies
- Apply cross-encoder and LLM-based rerankers where they earn their cost
- Decide when re-ranking adds quality vs only adds latency
6. Evaluation — retrieval and answer quality
- Measure retrieval quality with hit rate, MRR, and recall@k
- Measure answer quality with groundedness, faithfulness, and answer relevance
7. Common RAG failures and fixes
- Diagnose the 6 most common RAG failure modes from query traces
- Apply targeted fixes — chunking, query rewriting, metadata filters
Capstone: Capstone: Ship an evaluated RAG pipeline
- Design an end-to-end RAG pipeline with hybrid retrieval and re-ranking
- Document the eval harness and acceptance criteria for production cutover
Capstone deliverable: Every learner who completes this course produces «Your Evaluated RAG Pipeline» — a tangible artifact you take back to your organization.
Curriculum live · full chapter content rolling out through 2026.
The outline, learning objectives, references, and capstone deliverable are published. Full chapter content (video, narration, exercises) ships progressively. Get notified when each chapter goes live.
References & sources
Built on cited sources — not vibes.
Every course is researched fresh against vendor documentation, regulatory sources, and peer-reviewed work. Sources used in this course:
Azure AI Search Documentation
Microsoft Learn · Source link
Pinecone Learning Hub — RAG Patterns
Pinecone · Source link
Azure AI Foundry — RAG Reference Architectures
Microsoft Learn · Source link
Hugging Face — Embedding Models
Hugging Face · Source link
OWASP Top 10 for LLM Applications
OWASP Foundation · Source link
Course details
Track
Builder
Level
Intermediate
Audience
Technical practitioner
Function
IT & Engineering
Industry
Cross-Industry
Stack
Microsoft, Open-source, Stack-agnostic
Paired Gennoor Way phase
innovate, build
Format
video, hands-on, interactive
You finished the course. Now what?
From course to outcome.
Reading this course is step one. The next step is applying it where you work. Here's how Gennoor helps — without the deck, without the pitch.
Run this for your team
A 2-day workshop or virtual cohort for up to 25 of your people, with exercises run on your data and a 30-day adoption plan.
From $5k · 2 weeks · function-specific
Apply this to your data
A 4–6 week pilot that takes what you learned and ships a working system inside your environment. Fixed scope, fixed price, code transferred day one.
From $25k · 6 weeks · production-grade
Just want to talk?
Free 30-minute call. No deck, no pitch. We listen to your situation and tell you honestly what makes sense — even if it isn't us.
Free · no commitment · 30 minutes
Or just keep learning. We recommend next:
Just finished «RAG Architectures — Foundations». Want this to go further at your organization?