AI Learning Roadmap

AI Learning Roadmap

An AI learning roadmap is a staged path for building AI capability without drowning in disconnected courses: programming and software engineering, maths and statistics, machine learning, deep learning and LLMs, then production AI engineering.

Key points

  • Howell recommends starting with Python because most machine-learning infrastructure, libraries, examples, and jobs still assume the Python ecosystem [src-075].
  • Software-engineering fundamentals matter early: data structures, algorithms, system design, problem solving, and backend languages become more relevant as AI work moves toward production systems [src-075].
  • The math layer should be targeted rather than degree-like: statistics, linear algebra, and calculus are the practical foundation for understanding models [src-075].
  • The machine-learning layer should precede narrow generative-AI practice because AI is broader than LLMs, diffusion models, or ChatGPT-style applications [src-075].
  • The deep-learning/LLM layer should include PyTorch, neural networks, transformers, and practical LLM intuition before jumping into products and agents [src-075].
  • The final layer is AI Engineering Skill Stack: taking existing foundation models and shipping useful systems around them [src-075].
  • The roadmap is not a checklist to finish end-to-end. Howell's advice is to learn fundamentals from one resource, then apply them through projects [src-075].
  • Back to Engineering adds a physical-AI branch to the roadmap: electronics, microcontrollers, mechanical builds, ROS, edge compute, robot data loops, and embodied AI are a parallel path for applying AI to real-world systems [src-076].

Related entities

Related concepts

Source references

  • [src-075] Egor Howell — "STOP Taking Random AI Courses – Read These Books Instead" (2025-06-14)
  • [src-076] Back to Engineering (iulia) – physical AI, robotics, and data science cluster (41 videos, 2018-12-16 to 2026-05-10)