Resources and Bibliography

Cross-cutting references for the program, commented. Each module also has its own "For Further Reading" section at the end of each theory file.

Foundational Papers#

• Attention Is All You Need (Vaswani et al., 2017) — the transformer paper. Read it after the theory of module 1.
• Language Models are Few-Shot Learners (Brown et al., 2020, GPT-3) — origin of few-shot prompting.
• Chain-of-Thought Prompting Elicits Reasoning (Wei et al., 2022) and Self-Consistency (Wang et al., 2022) — foundation of module 2.
• Retrieval-Augmented Generation for Knowledge-Intensive NLP (Lewis et al., 2020) — the RAG paper.
• ReAct: Synergizing Reasoning and Acting (Yao et al., 2022) — the agent pattern you will implement by hand in module 4.
• Constitutional AI (Bai et al., Anthropic, 2022) — alignment, module 4.

Essential Official Documentation#

• Prompting guides from Anthropic (docs.claude.com) and OpenAI (platform.openai.com/docs) — the best practical guides available; reread them every few months.
• Building effective agents (Anthropic) — the reference essay on when to use agents and when not to.
• Model Context Protocol — modelcontextprotocol.io (spec and SDKs).
• LangGraph — langchain-ai.github.io/langgraph.
• RAGAS — docs.ragas.io.
• Amazon Bedrock — docs.aws.amazon.com/bedrock.
• OWASP Top 10 for LLM Applications — genai.owasp.org (module 5).

Free Complementary Courses#

• DeepLearning.AI short courses (deeplearning.ai/short-courses) — 1-2 hour nuggets on prompting, RAG, agents, evals; useful as reinforcement per module.
• AWS Skill Builder — official free path for AIF-C01.
• fast.ai Practical Deep Learning — if you need to reinforce the DL fundamentals prior to the program.

Newsletters and Tracking the State of the Art#

The model landscape changes every few months: rely on the release notes from OpenAI/Anthropic/Google/Meta, on LMSYS Chatbot Arena for live comparisons, and be skeptical of any static benchmark tables (including those in this repo if there ever were any).

How to Study This Program#

1. Theory → lab → exercises, in that order, for each topic. Do not move to the next topic without having executed and broken the lab (change parameters, provoke errors, measure).
2. Your own notebook (notas/ is in .gitignore if you want to create it): explaining a concept in your own words is the best test of comprehension.
3. Commits per topic: treat your progress as a real project; the git history is your study log.
4. Certification mock exams, cold: take them with a timer and without looking at notes, and then review the failed areas with the domain-specific guides.