Curated resources
A hand-picked reading list to go deeper than the demos. Start with whatever maps to the project you're on, then follow the foundational papers when you want the full picture.
Foundational papers
The primary sources behind modern LLMs and retrieval-augmented systems.
- Attention Is All You Need (Transformer)
The architecture every modern LLM is built on.
- Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
The paper that introduced RAG.
- Dense Passage Retrieval (DPR)
Embedding-based retrieval that powers modern RAG.
- ReAct: Synergizing Reasoning and Acting in Language Models
The reason+act loop behind tool-using agents.
- Chain-of-Thought Prompting
Step-by-step reasoning improves complex tasks.
- Toolformer: Language Models Can Teach Themselves to Use Tools
Models learning when and how to call tools.
- Self-RAG: Learning to Retrieve, Generate, and Critique
Self-reflective retrieval — the idea behind Agentic RAG.
Guides & documentation
Practical, well-maintained references you'll come back to.
- OpenAI Prompt Engineering Guide
Official, practical prompting patterns.
- Prompt Engineering Guide (DAIR.AI)
Comprehensive open guide to prompting techniques.
- LangChain — RAG concepts
How retrieval pipelines are structured in practice.
- LlamaIndex — Agentic RAG
Routing, query rewriting and tool-using retrieval.
- Hugging Face — Transformers course
Free, hands-on intro to the model stack.
- Anthropic — Building effective agents
When to use workflows vs. agents, with patterns.
Articles & tutorials
Approachable deep-dives and Medium-style write-ups to supplement each project.
- What is Retrieval-Augmented Generation?
Clear conceptual overview of RAG.
- A Visual Guide to Vector Embeddings
Build intuition for embeddings and similarity.
- Agentic RAG explained
Walkthrough of self-correcting retrieval loops.
- Multi-Agent LLM systems: patterns
Curated tag of multi-agent architecture write-ups.
- The Illustrated Transformer
The clearest visual explanation of attention.
- Patterns for Building LLM-based Systems & Products
Evaluation, RAG, caching and guardrails in production.