Foundations

What you'll learn

1. 1

   Your First Day as an AI EngineerFree

   Fix a real AI feature in your first five minutes, and see what the job actually is.

   • Day one: your first ticket
   • Find the bug in the prompt
   • Rewrite it
   • Never trust the reply
   • So what is an AI engineer?
   • Your next ticket

   Take-home template: AI Engineer 90-Day Plan & Skills Checklist

2. 2

   Tokens: What Models Actually ReadFree

   Models don't read words, they read tokens. Tokens decide your cost and your limits.

   • Text in, tokens out
   • Watch: budget an AI feature
   • See tokenization
   • Tokens and languages
   • Estimate the cost of a call
   • Context window math

   Take-home template: Token & Cost Calculator

3. 3

   Next-Token Prediction & TemperatureFree

   How an LLM picks each word, and the settings that control it.

   • One token at a time
   • Play with temperature
   • Implement softmax with temperature
   • Pick the right setting
   • Know your generation settings

   Take-home template: Generation Settings Cheat Sheet

4. 4

   How LLMs Are Built (and Why They Hallucinate)Free

   Pretraining, instruction tuning and preference tuning, and what each stage means for your product.

   • Three stages from raw text to assistant
   • Put the pipeline in order
   • Base vs instruct
   • Why models hallucinate
   • Measure how surprised a model is
   • Know the vocabulary

   Take-home template: Model Selection Scorecard

5. 5

   Context Windows, Latency & CostFree

   The three constraints behind every design decision in an AI product.

   • Bigger isn't always better
   • Predict response time
   • Cheaper or more expensive?
   • Where to put the important stuff

   Take-home template: Chat History Trimmer