Machine learning splits into two very different kinds of learning: the practical “get a model running” side and the theoretical “understand why it works” side. You need both eventually, so this list mixes them. (Before you start, the linear-algebra and calculus videos on my computer science list will save you a lot of pain.)
Practical and applied
Jeremy Howard: The fast.ai founder’s “Practical Deep Learning for Coders.” Famous for a top-down approach: build working models first, understand the theory as you go. A great place to start if math-first courses have burned you out before.
Abhishek Thakur: Applied ML from the world’s first quadruple Kaggle Grandmaster. Real frameworks, real competitions, practical tips.
Aladdin Persson: Hands-on PyTorch and TensorFlow tutorials plus clean implementations of core algorithms. The channel to follow while you are actually writing code.
Data School: Approachable machine learning and data analysis in Python with pandas and scikit-learn. Gentle and beginner-friendly.
Theory and deep learning
Ahlad Kumar: Deep, patient theory on neural networks and convolutional networks for when you want the math behind the layers.
Andreas Mueller: A scikit-learn core developer’s full “Applied Machine Learning” university course. Authoritative and well-structured.
Research and staying current
Yannic Kilcher: Clear walkthroughs of new ML papers across NLP, vision and architectures. The most efficient way to keep up without reading every paper yourself.
Henry AI Labs: Paper summaries and overviews of areas like reinforcement learning and GANs.
Rasa: “NLP for Developers” and practical work on building conversational AI assistants.
OpenAI: Talks, research releases and robotics work straight from the source.
One honest suggestion: pick Jeremy Howard’s course and finish it before you go hunting for more. The problem with ML learning is never a shortage of videos; it is jumping between them and never shipping a model.
Write a comment