Deep Learning with PyTorch: Zero to Research
Build and train neural networks from scratch: CNNs, Transformers, Diffusion Models, and RL agents.
4.8(4 reviews)5 studentsAdvancedEnglish
Created by Marcus Johnson
Data Science & AI
What you'll learn
- Implement neural networks from scratch in NumPy
- Master PyTorch 2.4 for production-grade DL
- Build and train Transformers and ViTs
- Understand and implement diffusion models
- Read and reproduce ML research papers
Requirements
- Solid Python and NumPy skills
- Linear algebra and calculus fundamentals
- Basic ML knowledge (regression, classification)
About this course
From first principles to cutting-edge research. This course goes beyond tutorials to build real understanding of how deep learning works. You'll implement neural networks from scratch in NumPy, then leverage PyTorch 2.4 for modern architectures. Topics: automatic differentiation, CNNs, ResNets, Transformers (from scratch), ViTs, diffusion models, reinforcement learning basics, and how to read ML papers.
Course curriculum
4 lectures · 320 min total
- Neural Networks from ScratchPreview60:00
- PyTorch Deep Dive70:00
- Convolutional Networks & Computer Vision90:00
- Transformers & Large Language Models100:00
Your instructor
Marcus holds a PhD in Computer Science from MIT with a focus on deep learning. He spent 4 years at Google Brain and has published 12 peer-reviewed papers. His mission is to demystify AI for working engineers and make cutting-edge ML techniques practical.
Student reviews
4.8- SSophia Bennett
Building a transformer from scratch is exactly what I needed to move from 'using AI tools' to 'understanding AI systems'. Marcus's intuition-first approach is world-class. Definitely my favorite course on this platform.
- BBenjamin Hughes
I have a statistics background and even I learned things I didn't expect in this course. The backpropagation chapter with visualizations is the best I've ever seen — beats any textbook I've used.
- HHenry Price
