AI Engineering: Hands-on provides a structured, hands-on curriculum for learning AI engineering from first principles. Targeting beginners and experienced engineers alike, it offers clear Jupyter notebooks to build intuition and practical skills in areas ranging from foundational math and PyTorch to modern LLM systems like transformers and RAG. The repository aims to equip users with the knowledge to construct real AI systems.

How It Works

This repository employs a sequential, notebook-driven approach, guiding users through core AI concepts. It begins with essential math fundamentals and PyTorch basics, progressing to building neural networks from scratch, implementing transformer architectures, and developing end-to-end Retrieval-Augmented Generation (RAG) and Optical Character Recognition (OCR) pipelines. This layered structure facilitates a deep, intuitive understanding of AI system components and their integration.

Quick Start & Requirements

• Install: Run pip install -r requirements.txt. Note that specific subfolders (e.g., RAG, OCR) may have additional dependencies listed in their own requirements.txt files.
• Prerequisites: Python environment with Jupyter (Lab or Notebook).
• Usage: Navigate to the project root and launch jupyter lab or jupyter notebook. Work through the notebooks sequentially following the recommended learning path (Start_here/learning_path.md).
• Links: Learning Path

Highlighted Details

• Covers foundational math (calculus, linear algebra, probability) and PyTorch tensor operations.
• Builds neural networks from scratch, including normalization techniques and optimizers.
• Implements transformer attention mechanisms and decoder-only architectures.
• Provides end-to-end RAG pipeline examples with indexing and retrieval strategies.
• Includes an OCR pipeline implementation.

Maintenance & Community

Contributions are welcomed with specific guidelines for notebook cleanliness and structure. The repository is marked for ongoing additions in the coming months. No specific community channels (like Discord/Slack) or roadmap links are provided.

Licensing & Compatibility

This project is licensed under the MIT License, which permits broad use, including commercial applications, with minimal restrictions.

Limitations & Caveats

Dependency management may require attention due to separate requirements.txt files in subdirectories. The repository focuses on practical implementation via notebooks and recommends external resources for deeper theoretical study. Ongoing development suggests potential for future changes.