Andrew Ng vs fast.ai vs Karpathy: Which ML Course Is Shortest?

Three courses dominate every “how do I learn machine learning?” discussion: Andrew Ng’s foundational course, Jeremy Howard’s fast.ai, and Andrej Karpathy’s Zero to Hero series. They’re all free. They’re all excellent. But they’re not the same size.

If you want to get building as fast as possible, which one do you pick? Here’s the data.

The three-way comparison

You can reproduce this comparison live by pasting all three playlist URLs into the compare tool.

Andrew Ng’s course: the foundation

Released in 2012 on Coursera, this is the course that created the modern ML practitioner ecosystem. It’s not the shortest path to building things, but it gives you the mathematical bedrock that makes everything else legible.

fast.ai Practical Deep Learning: the fastest path to state of the art

Jeremy Howard’s philosophy is explicit: start from working code and work backwards to theory. In practice, this means you’re fine-tuning a ResNet in week one and understanding why it works in week eight.

Karpathy’s Zero to Hero: the deepest understanding per hour

Karpathy builds everything from scratch — autograd, neural networks, transformers, GPT. It’s the smallest course by total duration and arguably the most dense per minute of content.

The verdict: which should you pick?

What to do after you finish