A complete, end-to-end curriculum covering deep learning fundamentals through LLM fine-tuning, RAG, and production deployment.

📖 Overview

Welcome to the Applied LLM Development repository! This project contains a comprehensive course structure designed to take you from the basics of PyTorch to operationalizing state-of-the-art Large Language Models (LLMs) in production.

All slide content from the original course has been transformed into interactive, self-contained Jupyter Notebooks. Each notebook combines rich markdown explanations with fully runnable code, giving you a hands-on learning experience.

Whether you're a data scientist, software engineer, or AI enthusiast, this curriculum provides the practical skills necessary to build, fine-tune, and deploy LLM applications.

---

✨ Key Features

• Interactive Learning: Theory and practice combined in Jupyter Notebooks.
• Progressive Curriculum: From basic PyTorch tensors to advanced Parameter-Efficient Fine-Tuning (PEFT) and LLMOps.
• Real-World Projects: Sentiment analysis, custom LLM fine-tuning, and RAG pipelines.
• Modern Tech Stack: PyTorch, Hugging Face Transformers, PEFT, QLoRA, and LangChain/LlamaIndex concepts.

---

🎯 What You Will Learn

• Deep Learning Fundamentals: Build neural networks from scratch using PyTorch.
• Computer Vision & NLP: Work with CNNs, RNNs, LSTMs, and sequence models.
• Transformer Architecture: Understand self-attention, positional encoding, and BERT.
• LLM Fine-Tuning: Master techniques like LoRA and QLoRA for efficient model adaptation.
• LLMOps: Deploy models robustly, track performance, and implement guardrails in production environments.

---

📂 Repository Structure & Curriculum

The repository is structured sequentially. We recommend progressing from Module 01 through Module 07.

Click to expand the full directory tree

Applied-LLM-Development/
│
├── 01-Deep-Learning-with-PyTorch-Introduction/
│   ├── notebooks/ (Intro to Deep Learning, Forward Pass, Activations, Data)
│   └── datasets/
│
├── 02-Deep-Learning-with-PyTorch-Intermediate/
│   ├── notebooks/ (PyTorch OOP, CNNs, RNNs/LSTMs, Multi-Input Models)
│   └── datasets/
│
├── 03-Deep-Learning-for-Text-with-PyTorch/
│   ├── notebooks/ (Text Preprocessing, Classification, Generation, BERT)
│   └── datasets/
│
├── 04-Introduction-to-LLMs-in-Python/
│   └── notebooks/ (Attention Mechanisms, Hugging Face APIs, Evaluation)
│
├── 05-PROJECT-Analyzing-Car-Reviews-with-LLMs/
│   └── Project_Analyzing_Car_Reviews_with_LLMs.ipynb
│
├── 06-PROJECT-Fine-Tuning-Your-Own-Llama2-Model/
│   └── Project_Fine_Tuning_LLaMA2.ipynb
│
├── 07-PROJECT-LLMOps-Operationalizing-LLMs/
│   └── Project_LLMOps_Guide.ipynb
│
├── Pytorch-cheatsheet/
│   └── (Quick reference guides and cheat sheets for PyTorch)
│
└── _img/
    └── (Assets and banners)

---

📚 Module Descriptions

🟢 Fundamentals

• Module 01: Deep Learning with PyTorch - Introduction
  Topics: Tensors, Forward Passes, Training Loops, Binary & Multiclass Classification, Activation Functions (Sigmoid, ReLU), Custom Datasets, and DataLoaders.
• Module 02: Deep Learning with PyTorch - Intermediate
  Topics: OOP in PyTorch, Image Handling, CNN architectures, Sequential Data, RNNs, LSTMs, Time Series Splitting, and Multi-input Models.
• Module 03: Deep Learning for Text with PyTorch
  Topics: NLP Pipelines, Tokenization, Word Embeddings, Padding, Character-level Text Generation, Transfer Learning, and BERT Fine-tuning.

🔵 Advanced & Applied LLMs

• Module 04: Introduction to LLMs in Python
  Topics: Attention Mechanisms, Positional Encoding, Self-Attention, Hugging Face pipeline() API, AutoModel, AutoTokenizer, and Evaluation Metrics (Accuracy, F1, BLEU, ROUGE, Perplexity).
• Module 05: Project - Analyzing Car Reviews with LLMs
  Topics: A full end-to-end pipeline covering loading data, sentiment analysis, summarization, visualization, and model evaluation.
• Module 06: Project - Fine-Tuning Your Own LLaMA Model
  Topics: Parameter-Efficient Fine-Tuning (PEFT), QLoRA, LoRA adapters, llama.cpp inference, and transitioning to modern architectures (LLaMA 2 / LLaMA 3).
• Module 07: Project - LLMOps & Operationalizing LLMs
  Topics: Prompt versioning, RAG (Retrieval-Augmented Generation) pipelines, performance monitoring, guardrails, and best practices for production deployments.

---

🛠️ Prerequisites

• Basic understanding of Python programming.
• Familiarity with fundamental math (linear algebra and calculus) is helpful but not strictly required.
• A machine with a GPU (or access to Google Colab/Kaggle) is highly recommended for Modules 04–07.

---

⚙️ Setup & Installation

To run the notebooks locally, we recommend using a virtual environment (such as conda or venv).

# 1. Clone this repository
git clone https://github.com/mohd-faizy/Applied-LLM-Development.git
cd Applied-LLM-Development

# 2. Create a virtual environment (Optional but recommended)
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# 3. Install core dependencies
pip install torch torchvision torchaudio
pip install transformers datasets evaluate
pip install peft trl bitsandbytes accelerate
pip install scikit-learn pandas matplotlib nltk torchtext jupyterlab

---

🚀 Getting Started

1. Open Jupyter Lab or Jupyter Notebook:

   jupyter lab

2. Navigate to 01-Deep-Learning-with-PyTorch-Introduction/notebooks/Chapter_01_Introduction_to_Deep_Learning.ipynb.
3. Work through the modules chronologically. Each notebook is self-contained with comprehensive markdown explanations and executable Python cells.

---

📎 Credits

• Course content based on the Developing Large Language Models curriculum.
• Notebooks generated and thoughtfully expanded from original course slides to provide a fully interactive learning environment.

---

Contributing and Support

Contributions are welcome. Please open an issue before submitting major changes.

If this repository helps you, consider giving it a star so other learners can discover it.

Connect with me

---

License

Distributed under the MIT License. See LICENSE for more information.