sota-ai-learning-roadmap

An expanded AI engineering study roadmap built from the source text of Chip Huyen's AI Engineering and rewritten into a much deeper self-study course with stronger explanations, mathematical foundations, production principles, and practical system-design context.

What this repository contains

• AI_Engineering_Course.md: the main long-form course, expanded into a structured roadmap with conceptual, mathematical, and production-level explanations.
• image.png: the repository cover image used in this README.
• pdf/ai_engineering.txt: the extracted source text used as the primary reference for the course rewrite.
• pdf/: additional AI, deep learning, NLP, and general AI reference texts.
• slugs/: supporting text assets included in the working directory.

Course highlights

• Covers the full modern AI engineering stack from foundation models to production systems.
• Expands each major section with deeper first-principles explanations and mathematical intuition.
• Adds practical framing around evaluation, retrieval, agents, finetuning, inference optimization, and feedback loops.
• Keeps the material beginner-friendly while still going deeper for serious learners.

Module map

1. Module 0: Setting the stage for AI engineering
2. Module 1: Introduction to AI engineering
3. Module 2: Understanding foundation models
4. Module 3: Evaluation methodology
5. Module 4: Evaluating AI systems
6. Module 5: Prompt engineering
7. Module 6: RAG and agents
8. Module 7: Finetuning
9. Module 8: Dataset engineering
10. Module 9: Inference optimization
11. Module 10: Architecture and user feedback

What makes this version different

This repository is not just a transcription or short summary. The course file was substantially enriched to include:

• root-level explanations of key concepts such as probability modeling, attention, scaling laws, retrieval, and optimization
• production-minded guidance on latency, cost, guardrails, routing, and monitoring
• deeper coverage of evaluation methods, including uncertainty, pairwise ranking, and rubric design
• stronger explanations of LoRA, quantization, RAG, agent loops, data quality, and deployment trade-offs

How to use this roadmap

Read AI_Engineering_Course.md from top to bottom for a full learning path, or jump directly to the module most relevant to your current focus area. The course is especially useful for:

• software engineers moving into AI engineering
• ML engineers shifting toward foundation model applications
• founders and builders designing production AI systems
• self-learners who want both intuition and technical depth

Source note

The main course in this repository is based on pdf/ai_engineering.txt, which was extracted from the original book source and then reworked into a richer markdown learning document.