Skip to content

Complete AI/ML Learning Guide

A structured, end-to-end roadmap for learning Artificial Intelligence, Machine Learning, Deep Learning, Generative AI, and Agentic Systems. This guide integrates content from this repository with external resources for comprehensive coverage.

Phase 0: Prerequisites

Build programming and tooling fundamentals required before entering AI/ML.

0.1 Python for AI/ML

Python basics and essential data-science libraries. See the coding environment setup for local development configuration.

Environment Setup with uv

uv is the recommended package manager for this repository. It provides fast dependency resolution and virtual environment management.

# Install uv
curl -LsSf https://astral.sh/uv/install.sh | sh

# Initialize project and sync dependencies
uv sync

# Run scripts
uv run python script.py

Key Libraries

Library Description Documentation
NumPy Array computing and numerical operations coding/index.md#numpy
Pandas Data manipulation and analysis pandas.md
Matplotlib Data visualization visualization.md
Seaborn Statistical graphics coding/index.md#seaborn
PyTorch Pytorch library coding/pytorch.md

External Resources:

0.2 Git Basics

Version control is required for all ML and AI work.

External Resources:

0.3 Linux Commands (Optional)

Useful for development environments and servers.

External Resources:

Phase 1: Mathematical Foundations

Understand why ML and DL models work, not just how to use them.

1.1 Linear Algebra (Core of ML)

Core concepts: Scalars, vectors, matrices, tensors; vector operations (dot product, cross product, norm); matrix operations (multiplication, transpose, inverse); eigenvalues and eigenvectors; PCA intuition.

Used in: Neural Networks, Linear Regression, PCA, Embeddings, Attention, Transformers.

External Resources:

1.2 Probability and Statistics (ML Reasoning)

See Mathematical Foundations for:

  • Probability basics and conditional probability
  • Bayes theorem (prior, likelihood, posterior)
  • Data distributions (Gaussian, Poisson, Uniform)
  • Covariance and correlation
  • Normalization techniques

Key Notebooks:

External Resources:

1.3 Calculus (Optimization and Learning)

See ML Concepts - Cost Function for:

  • Gradient and direction of steepest descent
  • Gradient descent algorithm
  • Learning rate and convergence

External Resources:

1.4 Bias-Variance Tradeoff

See ML Concepts for detailed coverage of:

  • Variance and model consistency
  • Bias and prediction accuracy
  • Regularization (L1/Lasso, L2/Ridge, Elastic Net)
  • Overfitting and underfitting

Phase 2: Core Machine Learning

Learn classical ML using feature-based models and structured data.

2.1 Introduction to Machine Learning

See Machine Learning Overview for:

  • ML vs AI vs Deep Learning
  • Supervised, unsupervised, reinforcement learning
  • Classification vs regression
  • ML workflow and system design

2.2 Data Understanding and Preprocessing

See Feature Engineering for:

  • Handling missing values
  • Categorical encoding (ordinal, one-hot)
  • Feature scaling and normalization
  • Mutual information for feature selection
  • Creating new features

2.3 Supervised Learning - Regression

See ML Index - Regression for:

  • Linear, multiple, and polynomial regression
  • Hypothesis functions
  • Cost functions (MSE)
  • Gradient descent

Metrics: MAE, MSE, RMSE, R-squared. See Performance Metrics.

2.4 Supervised Learning - Classification

See Classifiers for detailed implementations:

Algorithm Description Code
Perceptron Basic neural unit TestPerceptron.py
Adaline Adaptive Linear Neuron TestAdaline.py
Logistic Regression Probability-based classification classifier.md#logistic-regression
SVM Maximum margin classification SVM-IRIS.py
Decision Trees Rule-based learning DecisionTreeIRIS.py
Random Forest Ensemble learning classifier.md#random-forests
KNN Instance-based learning KNN Notebook

Metrics: Confusion matrix, accuracy, precision, recall, F1, ROC-AUC.

2.5 Unsupervised Learning

See Unsupervised Learning for:

  • K-Means clustering
  • Cluster labels and distance features
  • Dimensionality reduction

2.6 Model Selection and Validation

See ML System for:

  • Train/validation/test split
  • Cross-validation (K-fold, LOOCV)
  • Bias-variance tradeoff in practice

2.7 ML Libraries (Hands-On)

See Coding Index for environment setup and:

External Resources:

Phase 3: Deep Learning and Advanced ML

Build and understand neural-network-based systems end-to-end.

3.1 Neural Network Fundamentals

See Deep Learning for:

  • Neuron structure and activation
  • Input, hidden, and output layers
  • Activation functions (Sigmoid, ReLU, Softmax)
  • Forward and backward propagation

3.2 PyTorch Framework

See PyTorch for comprehensive coverage:

  • Tensors and GPU computation
  • Neural network modules (torch.nn)
  • Optimizers and loss functions
  • Training workflows

Key Notebooks:

3.3 Classification Neural Networks

See Classification Architecture for:

  • Layer design and hyperparameters
  • Loss functions (Cross entropy, BCE)
  • Optimizer selection (SGD, Adam)

3.4 Convolutional Neural Networks (CNNs)

See CNN Section for:

  • Convolution and pooling layers
  • Image processing architecture
  • Feature extraction

Code Examples:

3.5 Transfer Learning

See Transfer Learning for:

  • Pre-trained models usage
  • Fine-tuning strategies
  • Feature freezing

External Resources:

Phase 4: LLMs, NLP and Generative AI

Use transformer-based LLMs to build real-world AI applications.

4.1 Generative AI Foundations

See Generative AI Overview for:

  • Transformer architecture
  • Pre-training and fine-tuning
  • NLP processing and tokenization
  • Embeddings and context windows

4.2 LLM Fundamentals

See GenAI Concepts for:

  • Encoder-decoder architectures
  • Inference parameters (Temperature, Top-K, Top-P)
  • Model selection considerations

4.3 Prompt Engineering

See Prompt Engineering for:

  • Zero-shot and few-shot prompting
  • Chain of Thought (CoT)
  • Prompt chaining
  • Tree of Thoughts
  • Automatic Prompt Engineering

4.4 Embeddings and Vector Databases

See GenAI - Vector Database and NLP Embeddings for:

  • Similarity search
  • ChromaDB, FAISS, OpenSearch
  • Embedding models

4.5 Retrieval-Augmented Generation (RAG)

See RAG for comprehensive coverage:

  • RAG architecture (indexing, retrieval, generation)
  • Document pipelines and chunking
  • Retriever considerations
  • Advanced RAG techniques (multi-query, RAG fusion, HyDE)
  • Knowledge graph integration

4.6 LLM Providers

Provider Documentation
OpenAI openai.md
Anthropic Claude anthropic.md
Mistral mistral.md
Cohere cohere.md

4.7 LLM Development Frameworks

Framework Documentation
LangChain langchain.md
LlamaIndex llama-index.md
Haystack haystack.md

External Resources:

Phase 5: Agentic Systems and AI System Design

Design autonomous, goal-driven AI systems with tools, memory, orchestration, and safety controls.

5.1 From LLMs to Agents

See Agentic AI for:

  • Agent reference architecture
  • Planning strategies (CoT, Tree of Thoughts, ReAct)
  • Memory systems (short-term, long-term, entity)
  • Tool integration

5.2 Agent Design Patterns

See Agentic Guidelines for:

  • Role definition and focus
  • Tool selection and management
  • Multi-agent cooperation
  • Guardrails and control

5.3 LangGraph for Agent Orchestration

See LangGraph for:

  • Stateful multi-actor applications
  • Graph-based workflows
  • Conditional edges and routing
  • Human-in-the-loop patterns
  • Persistence and checkpointing

Key Patterns:

5.4 Multi-Agent Frameworks

Framework Description Documentation
LangGraph Graph-based orchestration langgraph.md
CrewAI Multi-agent collaboration agentic.md#crewai
AutoGen Conversable agents agentic.md#autogen
OpenSSA Small Specialist Agents agentic.md#openssa

5.5 Model Context Protocol (MCP)

See MCP for:

  • Standardized tool integration
  • Context management
  • Protocol implementation

5.6 Agent Use Cases

See Agentic Use Cases for examples:

  • Research and writing agents
  • Customer support crews
  • Sales lead analysis
  • Job application tailoring

Code Examples:

External Resources:

Supporting Topics

UI Frameworks for AI Applications

Framework Documentation
Streamlit streamlit.md
Gradio gradio/index.md
NiceGUI nicegui.md
Taipy taipy/index.md

Cloud and Infrastructure

Platform Documentation
GCP gcp/index.md
Feature Stores feature_store.md
Airflow airflow.md

Methodology

See Methodology for project planning approaches.

Books and Resources

See the main index for a comprehensive list of books and resources including:

  • Python Machine Learning - Sebastian Raschka
  • Collective Intelligence - Toby Segaran
  • Stanford ML Course - Andrew Ng
  • Dive into Deep Learning
  • Kaggle competitions
  • Papers with Code

Learning Path Recommendations

Beginner Path (2-3 months)

  1. Phase 0: Python basics, NumPy, Pandas
  2. Phase 1: Linear algebra essentials, probability basics
  3. Phase 2: Scikit-learn classifiers, regression basics

Intermediate Path (3-4 months)

  1. Phase 3: PyTorch fundamentals, neural networks
  2. Phase 3: CNNs and transfer learning
  3. Phase 4: LLM basics, prompt engineering

Advanced Path (3-4 months)

  1. Phase 4: RAG implementation, fine-tuning
  2. Phase 5: LangGraph agents
  3. Phase 5: Multi-agent systems, production deployment

Notes

  • This guide prioritizes practical implementation with code examples from this repository.
  • External video resources are provided for topics requiring deeper theoretical understanding.