🤖 AI Foundations Lab

A professional collection of AI/ML fundamentals, implemented from first principles to advanced applied projects.

Explore Projects • Key Features • Connect

---

⭐ Star this repo if you find it useful!

📌 Table of Contents

• 📖 About
• ✅ Key Features
• 🚀 Project Showcase
• 🖼️ Showcase Gallery
• 🛠️ Technologies
• 📂 Repository Structure
• 🗺️ Roadmap / Future Work
• 🏁 Getting Started
• 📚 Learning Resources
• 🤝 Connect With Me

---

📖 About

This repository serves as a comprehensive laboratory for my journey through the foundations of Artificial Intelligence. Driven by a "Back to Basics" and "First Principles" philosophy, this project documents my transition from academic theory (Stanford/MIT curriculum) to high-fidelity implementation.

Whether it's deriving the gradients for Logistic Regression or architecting a Deep Convolutional GAN, every line of code here is built with rigor, documentation, and a focus on visual performance analytics.

---

✅ Key Features

• ✅ From-scratch implementations: Neural networks and ML algorithms built using only NumPy.
• ✅ LaTeX math explanations: Detailed derivations for cost functions, loss functions, and optimization algorithms.
• ✅ Clean documented code: High-quality, professional code with comprehensive comments.
• ✅ Jupyter visualizations: Real-time plotting of training curves, decision boundaries, and model outputs.
• ✅ Real-world projects: Applied computer vision (YOLOv8) and generative models (GANs).

---

🚀 Project Showcase

🤖 Machine Learning

Focuses on classical algorithms, statistical learning theory, and supervised/unsupervised learning fundamentals.

• Includes: Linear/Logistic Regression, K-Means, PCA, Decision Trees.

🧠 Deep Learning

Deep dives into multi-layer perceptrons, convolutional neural networks (CNNs), and optimization techniques.

• Includes: Backpropagation, CNN Architectures, Regularization, Batch Norm.

💬 NLP

exploration of sequences and linguistics in AI.

• Includes: Word Embeddings (Word2Vec), RNNs, LSTMs, and an introduction to Transformers.

🎮 Reinforcement Learning

Implementing agents that learn from interaction with their environment.

• Includes: Q-Learning, SARSA, and Deep Q-Networks (DQN).

⭐ Applied Projects

This is where the theory meets the real world. This section highlights high-fidelity implementations of state-of-the-art architectures.

[NEW] Object Detection dengan YOLOv8

Complete inference and training pipeline for YOLOv8, including hardware acceleration (CUDA) and performance analytics.

[NEW] Generative Adversarial Networks (GANs)

Implementation of DCGAN for synthetic image generation, exploring adversarial loss and transposed convolutions.

---

🖼️ Showcase Gallery

🎯 Object Detection (YOLOv8)

[!NOTE] > Caption: Real-time object detection with YOLOv8 - detecting people, vehicles, and objects.

Detecting complex scenes with high confidence counts.

Human and tie detection on sample imagery.

Real-time video inference frame analysis.

🎨 Generative Art (GANs)

[!NOTE] > Caption: Fashion-MNIST: Real vs Synthetic images generated by GAN.

Synthetic fashion items generated after adversarial training.

---

🛠️ Technologies

---

📂 Repository Structure

ai-foundations-lab/
├── 01-machine-learning/          # Classical ML algorithms
├── 02-deep-learning/             # Neural Networks & CNNs
├── 03-nlp/                       # Natural Language Processing
├── 04-reinforcement-learning/    # RL Agents & Environments
├── 05-applied-projects/          # SOTA Architectures (YOLO, GANs)
│   ├── Generative Adversarial Networks/
│   └── Object-Detection-YOLO/
├── assets/                       # Repository images & media
└── requirements.txt              # Project dependencies

---

🗺️ Roadmap / Future Work

• Vision Transformers (ViT): Implementation and comparison with CNNs.
• More RL environments: Training agents on Mujoco/Gymnasium environments.
• LLM fine-tuning: Experiments with LoRA and QLoRA on open-source LLMs.
• Web app deployments: Serving models via FastAPI and Docker.

---

🏁 Getting Started

1. Clone the repository:

   git clone https://github.com/DaviBonetto/ai-foundations-lab.git
   cd ai-foundations-lab

2. Set up the environment:

   python -m venv .venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate
   pip install -r requirements.txt

3. Explore the notebooks: Launch Jupyter and dive into any module!

---

📚 Learning Resources

This project is inspired by and follows the academic rigor of graduate-level computer science courses:

Stanford University:

• CS229: Machine Learning - Supervised learning, unsupervised learning, learning theory
• CS230: Deep Learning - Neural networks, CNNs, RNNs, optimization techniques
• CS224N: Natural Language Processing with Deep Learning - Word embeddings, transformers, attention mechanisms
• CS234: Reinforcement Learning - MDPs, Q-learning, policy gradients, deep RL
• CS221: Artificial Intelligence: Principles and Techniques - Search, reasoning, learning
• CS336: Language Models - Transformer architecture, LLM foundations

Additional Resources:

• MIT 6.S191: Introduction to Deep Learning
• Research papers and textbooks by leading AI researchers

Note: This is a self-directed learning project using publicly available course materials. No formal enrollment or academic credit.

---

🤝 Connect With Me

"The only way to learn a new programming language is by writing programs in it." — Dennis Ritchie

---

Developed by Davi Bonetto