🚀 Deep Learning Mastery — Foundations to Architectures

A professional-grade deep learning library featuring from-scratch implementations (NumPy) and state-of-the-art architectures (PyTorch) across Computer Vision, NLP, and Generative AI.

---

✨ Key Features

• 🏗️ From-Scratch Foundations — MLP, Optimizers (Adam), and Attention implemented in pure NumPy.
• 📸 Computer Vision — ResNet architectures and CNN projects (Face Recognition, VGG16).
• 🧠 NLP & Sequences — Scaled Dot-Product Attention and RNN/LSTM modules.
• 🎨 Generative Models — Variational Autoencoders (VAE) and GAN foundations.
• 🛠️ Production Best Practices — Batch Norm, Dropout, and professional training loops.

---

📂 Project Structure

Deep-Learning/
├── foundations/         # Pure NumPy (MLP, Adam, Activations)
├── vision/              # PyTorch CV (ResNet, CNN blocks)
├── nlp/                 # NLP & Attention mechanisms
├── generative/          # VAE and GAN implementations
├── projects/            # Full-scale Jupyter Notebook projects
│   └── notebooks/       # Face Rec, Medical Diagnosis, Hate Speech
└── README.md

---

🧠 Master Concept Checklist

Category	Concept	Implementation	Status
Foundations	MLP from Scratch	foundations/mlp_from_scratch.py	✅
Foundations	Adam Optimizer	foundations/optimizers_from_scratch.py	✅
Computer Vision	ResNet / Skip-Conn	vision/resnet_pytorch.py	✅
NLP	Attention Mechanism	nlp/attention_from_scratch.py	✅
Generative	VAE (PyTorch)	generative/vae_pytorch.py	✅

---

⚙️ Installation & Setup

# Clone the repository
git clone https://github.com/Vaishnavi-Dubey/Deep-Learning.git
cd Deep-Learning

# Install dependencies
pip install torch torchvision numpy matplotlib

---

📈 Impact / Highlights

• 🎯 Math to Code — Every from-scratch file includes derivations and complex math explained in comments.
• ⚡ Framework Proficiency — Balanced mix of low-level NumPy and high-level PyTorch modules.
• 📊 Scalable Reference — Structured to serve as a deep learning reference library for researchers and engineers.

---

📜 License

This project is licensed under the MIT License.

---

Built with ❤️ by Vaishnavi Dubey