LLM Fundamentals & Solution Choices

By the end, you can explain what an LLM is and how it turns context into output, and you can choose and justify the right solution pattern (prompting, RAG, fine-tuning, workflow or agent) for a business problem.

≈ 6 h 38 min Enterprise Bank Policy Assistant · v0.1 10 lessons · 10 playable now

0 of 10 lessons complete

Progress and your code save automatically

Every concept lesson has the same rhythm

WatchExploreRead or listenTry itBuildCheckSay it to a client

0
Start here: Northgate's challenge
State the Northgate brief, write your Day 1 design note, and get a working repo on the mock provider.
26 min·ListenReadWriteSet up·L0.1 Write your Day 1 design note · repo setup
1
What an LLM actually is New
Explain what a model is, how pre-training and instruction tuning shape it, what an API request and response contain, and how model families differ.
48 min·WatchExploreReadSlidesTry itWalkthroughBuildCheck·L1.1 Your first model call
2
Tokens and the context window
Explain tokens, the context window and why cost grows with input + output tokens; log the cost of every call.
54 min·WatchExploreReadDemoTry itBuildCheck·L2.1 Log every call
3
How a model writes: generation and temperature New
Explain next-token sampling and temperature, and why the app must supply conversation state.
25 min·WatchExploreReadLive demoTry itCheck
4
What a model knows (and makes up)
Explain hallucination, ground answers in a supplied policy, and make the bot refuse when the answer isn't there.
1 h 11 min·WatchExploreReadLive demoTry itBuildCheck·L4.1 Write the assistant's rules · L4.2 Give it the policy · L4.3 Prove it on a live model
5
Choosing a model
Compare two models on quality, latency, context size and cost, and justify a choice with data.
1 h 23 min·WatchExploreReadSlidesLive demoTry itBuildCheck·L5.1 Compare two models · L5.2 Record your model decision
6
The solution-pattern ladder
Explain how prompting, RAG, workflows, agents and fine-tuning differ, and when each fits.
23 min·WatchExploreReadSlidesTry itCheck
7
Debug the broken assistant
Diagnose and fix context, temperature and cost faults from evidence.
33 min·Break labCheck·L7.1 Find and fix what's wrong · L7.2 Prove the fix on a live model · L7.3 Write up your diagnosis
8
Pick the right pattern
Choose and justify a pattern for five Northgate requests plus the Day 1 assistant.
20 min·DecideCheck·Submit your six decisions
9
Module quiz
Pass the 10-question scenario quiz (7/10).
15 min·Quiz·Quiz score
Module completion checklistProve: the evidence that you've finished Module 1Open