Stanford LLM Curriculum: From Prompt User to AI Engineer
Large Language Models (LLMs) are the foundation of modern AI systems such as ChatGPT, Claude, and many other advanced AI tools.
But most people only interact with these models as users.
If you want to go deeper and understand how these systems actually work, you need to learn the concepts behind:
Transformers
LLM training
Instruction tuning
AI reasoning
Agent systems
Model evaluation
This guide follows the structure of Stanford’s LLM curriculum and organizes the lectures into a clear step-by-step learning path.
By studying these modules in order, you can move from prompt user → AI engineer.
How to Use This Guide
Follow this study process while going through the lectures.
• Watch lectures in the given order
• Take handwritten or digital notes
• After each lecture, summarize the key ideas in your own words
• Try to implement at least one small experiment per section
• Do not skip steps — each concept builds on the previous one
Module 1: Transformers Fundamentals
What you’ll learn
What tokens are
Embeddings and vector representations
Attention mechanism
Self-attention
Encoder vs decoder architectures
Why transformers replaced RNNs and CNNs in NLP
Outcome
You’ll understand the core architecture behind modern AI systems.
Watch
Transformers Fundamentals
https://youtu.be/Ub3GoFaUcds?si=h_-8aszab0WLDlG8
Module 2: Transformer Tricks & Optimization
What you’ll learn
Positional encoding
Layer normalization
Residual connections
Scaling strategies
Training stability techniques
Outcome
You’ll understand how transformers scale to billions of parameters.
Watch
Transformer Tricks
https://youtu.be/yT84Y5zCnaA?si=NKDHfaRNw8KEK0P1
Module 3: From Transformers to Large Language Models
What you’ll learn
What makes an LLM different from a small model
Scaling laws
Emergent abilities
Pretraining objectives
Data distribution effects
Outcome
You’ll understand why large models behave differently from smaller ones.
Watch
Large Language Models
https://youtu.be/Q5baLehv5So?si=fqxbQrwyRFhrYAWX
Module 4: LLM Pretraining
What you’ll learn
Pretraining pipelines
Token prediction objectives
Dataset construction
Compute and scaling considerations
Where “intelligence” emerges
Outcome
You’ll understand how base models are created and trained.
Watch
LLM Training
https://youtu.be/VlA_jt_3Qc4?si=GcKuwAWWiAdJ5PWO
Module 5: Instruction Tuning & Alignment
What you’ll learn
Supervised fine-tuning (SFT)
RLHF (Reinforcement Learning from Human Feedback)
PPO and DPO
Alignment techniques
Why raw models behave differently from chat models
Outcome
You’ll understand how base models become helpful AI assistants.
Watch
Module 6: LLM Reasoning
What you’ll learn
Why models make reasoning mistakes
Chain-of-thought prompting
Reinforcement learning for reasoning
GRPO and scaling reasoning
Outcome
You’ll understand how reasoning improves through training and prompting.
Watch
LLM Reasoning
https://youtu.be/k5Fh-UgTuCo?si=ookqU35nxwfEp9pJ
Module 7: Agentic LLMs
What you’ll learn
Tool calling
Retrieval-Augmented Generation (RAG)
Planning agents
Memory systems
Autonomous workflows
Outcome
You’ll learn how LLMs evolve from text generators to action-taking systems.
Watch
Agentic LLMs
https://youtu.be/h-7S6HNq0Vg?si=b3i2zKiZi7MgE6Gk
Module 8: LLM Evaluation
What you’ll learn
Benchmarking methods
LLM-as-a-judge
Evaluation datasets
Measuring reasoning and reliability
Why demos can be misleading
Outcome
You’ll learn how to properly evaluate AI systems.
Watch
LLM Evaluation
https://youtu.be/8fNP4N46RRo?si=wShichCLbslEECDb
Module 9: What’s Next in LLMs
What you’ll learn
Future trends in AI
Multimodal models
Better reasoning systems
More efficient architectures
The direction of AI research
Outcome
You’ll understand where the future of AI and LLMs is heading.
Suggested 14-Day Study Plan
Day 1–2: Transformers fundamentals
Day 3: Transformer tricks
Day 4–5: Large language models
Day 6–7: LLM pretraining
Day 8–9: Instruction tuning & alignment
Day 10: LLM reasoning
Day 11–12: Agentic LLMs
Day 13: LLM evaluation
Day 14: Future trends and revision
Final Advice
To truly understand LLM systems:
• Rewatch difficult lectures
• Write summaries after each module
• Build small projects while learning
• Focus on concepts, not just tools
Master these ideas once, and every AI tool you use afterwards will make much more sense.