naveenopslearn
Search

Learning path

AI Infrastructure

Start with how a computer works, follow a request across the Internet to a server, then meet the GPU and see why AI needs it. Short lessons, plain words, one diagram each.

Who it is for

Anyone curious about what really runs AI. You do not need a technical background: if you can use a phone or a laptop, you are ready.

Prerequisites

None. Every new word is explained the first time it shows up.

Total time

About 122 minutes of reading across 28 short lessons. 16 lessons (53 min) are live now; the rest are on the way.

Start lesson 1
Your progress16 lessons available

Course outline

  1. Module 1Ready

    How a computer works

    Input, processing and output; the CPU, memory and storage; and what happens when you press the power button.

    1. 01How a computer works
      1. 1.1What is a computer?(completed)3 min
      2. 1.2The CPU: the brain(completed)3 min
      3. 1.3Memory: RAM vs storage(completed)3 min
      4. 1.4How the CPU and memory work together(completed)3 min
      5. 1.5What happens when you turn on a computer(completed)4 min
  2. Module 2Ready

    How the Internet works

    Addresses, names and packets, and what a busy server does with many requests at once.

    1. 02How the Internet works
      1. 2.1What is the Internet?(completed)3 min
      2. 2.2Addresses and names: IP and DNS(completed)3 min
      3. 2.3Packets: how information travels(completed)3 min
      4. 2.4Client and server: opening a website(completed)3 min
      5. 2.5Inside a busy server(completed)4 min
      6. 2.6Why this matters for AI(completed)3 min
  3. Module 3Ready

    Introducing the GPU

    What a GPU is, how it differs from a CPU, its own memory, and why it suits AI.

    1. 03Introducing the GPU
      1. 3.1What is a GPU?(completed)3 min
      2. 3.2CPU vs GPU(completed)3 min
      3. 3.3The GPU's own memory: VRAM(completed)4 min
      4. 3.4How the CPU and GPU work together(completed)4 min
      5. 3.5Why GPUs suit AI(completed)4 min
  4. Module 4Coming soon

    Tensors and models

    The numbers AI works with, and what a "model" actually is.

    1. 04Tensors and models
      1. 4.1What is a tensor?soon5 min
      2. 4.2What is an AI model?soon5 min
      3. 4.3Training vs inferencesoon5 min
  5. Module 5Coming soon

    Splitting work across GPUs

    Data, tensor, pipeline and expert parallelism, and which to use when.

    1. 05Splitting work across GPUs
      1. 5.1Why split work across GPUs?soon5 min
      2. 5.2Data parallelismsoon6 min
      3. 5.3Tensor parallelismsoon6 min
      4. 5.4Pipeline parallelismsoon6 min
      5. 5.5Expert parallelismsoon6 min
      6. 5.6Which to use whensoon6 min
  6. Module 6Coming soon

    Serving LLMs with vLLM

    Turning a model into a service that answers many people at once.

    1. 06Serving LLMs with vLLM
      1. 6.1What is LLM serving?soon5 min
      2. 6.2Meet vLLMsoon6 min
      3. 6.3Serving a model with vLLMsoon8 min