These AI and machine learning videos come from HackerTube, Andrej Karpathy’s weekend-built index of YouTube links found in Hacker News comments. This independent guide turns the broad AI and data catalog into a smaller map of subtopics and a practical six-video starting route.
Start with machine learning and neural-network foundations
New to the field? Begin with the learning problem, then use a visual neural-network explanation before moving into gradients and backpropagation. That order separates the goal of machine learning from the mechanics of training a model.
Move from visual intuition to working code
Once layers and gradients make sense, follow the path into transformers, tokenization and a small GPT built from scratch. The featured sequence pairs visual explanations with Andrej Karpathy’s code-first teaching, so you can connect model diagrams to the operations behind them.
Treat Hacker News signals as context
HackerTube reflects what people linked and recommended in Hacker News discussions; it is not a universal ranking of AI courses. This hub favors source records labeled evergreen with positive recommendation history, but you should still check the summary, level and prerequisites before committing to a long lecture.
The ten-video path starts with the learning problem, adds visual intuition for backpropagation, then moves into micrograd, transformers, tokenization and a small GPT build.
Follow the AI foundations watch path ↗Six videos to start with
-
↗course · intro
Lecture 01 - The Learning Problem
A math-aware introduction that clarifies the learning problem before model details take over.
5 positive HN recommenders across 5 recommendation months
-
↗explainer · intro
But what is a neural network? | Deep learning chapter 1
A visual first pass through neurons, layers and the mathematics of a basic network.
16 positive HN recommenders across 14 recommendation months
-
↗lecture · intro
The spelled-out intro to neural networks and backpropagation: building micrograd
Karpathy builds a small autograd engine to make gradients and backpropagation concrete.
17 positive HN recommenders across 16 recommendation months
-
↗explainer · intro
Transformers, the tech behind LLMs | Deep Learning Chapter 5
A visual bridge from tokens and embeddings to attention and transformer behavior.
13 positive HN recommenders across 10 recommendation months
-
↗lecture · intermediate
Let's build the GPT Tokenizer
Karpathy exposes how byte pair encoding and Unicode shape the inputs an LLM sees.
5 positive HN recommenders across 5 recommendation months
-
↗tutorial · intermediate
Let's build GPT: from scratch, in code, spelled out.
A code-first route through constructing a small GPT and connecting the pieces end to end.
14 positive HN recommenders across 8 recommendation months
Narrow the topic
These narrower categories open the video directory with the matching filter already applied.