AI and ML Learning Resources

A curated collection of GitHub repositories for learning Artificial Intelligence, Machine Learning, and Machine Vision, categorized by sub-field and difficulty.

Overview

This collection tracks high-quality educational content, reference implementations, and industry-standard libraries.

Source: [[sources/github_repos_ml_research_2026.md]]

---

Machine Learning (General & Classical)

Focuses on foundational algorithms, statistical modeling, and non-deep learning techniques.

• Beginner: microsoft/ML-For-Beginners
  • 12-week curriculum for foundations (Scikit-learn, regression, clustering).
• Intermediate: trekhleb/homemade-machine-learning
  • Mathematical foundations by implementing classical algorithms from scratch in Python.
• Advanced: scikit-learn/scikit-learn
  • The industry standard for classical ML. Reading the source code is a masterclass in performant data science engineering.

Deep Learning & LLMs

The architecture and mechanics behind modern Generative AI.

• Beginner: fastai/fastbook
  • Top-down entry into deep learning utilizing the Fast.ai library.
• Intermediate: rasbt/LLMs-from-scratch
  • Step-by-step implementation of a GPT-style Large Language Model in PyTorch.
• Advanced: karpathy/nn-zero-to-hero
  • The definitive guide to bottom-up construction (autograd from scratch).
• Industry Standard: pytorch/pytorch
  • The core framework used in research and production.

Computer Vision (CV)

Image processing, object detection, and visual understanding.

• Beginner: spmallick/learnopencv
  • Tutorials spanning classical image processing to modern YOLO models.
• Intermediate: opencv/opencv
  • The absolute standard for classical CV.
• Advanced: facebookresearch/detectron2 & open-mmlab/mmdetection
  • Modular toolboxes for state-of-the-art object detection and instance segmentation.

MLOps & Production System Design

Bridging the gap between model code and production systems.

• Beginner: graviraja/MLOps-Basics
  • Weekly introduction to core tools (DVC, Hydra, ONNX).
• Intermediate: GokuMohandas/Made-With-ML
  • Emphasizes CI/CD, testing, and scaling for ML systems.
• Advanced: DataTalksClub/mlops-zoomcamp
  • Thorough coverage of orchestration (Mage/Prefect) and experiment tracking.

---

Learning Philosophies

• Top-Down: (e.g., fastai) Start with the API, get results, then drill down.
• Bottom-Up: (e.g., karpathy) Build every component from scratch to understand the mechanics.

Related: [[Learning Path - ML Development]], [[Syllabus]]