Command Palette

Search for a command to run...

Hectal
Phase 0Beginner1 of 15 in LLM Engineering

Start Here: What an LLM Is and Why Engineering Matters

What a large language model really does, why calling it from a demo is easy and shipping it is hard, and how this course runs code in three languages.

This phase assumes you can write a few lines in one programming language (Hectal's Python course is enough) and nothing about AI. It starts with what a large language model actually is: a program that continues text one small piece at a time. You will build a tiny one by hand from word counts and watch it write sentences.

Then it turns to engineering. A model call that works once in a demo fails in a dozen ways in a product: it is slow, it costs money, it gets rate-limited, it returns the wrong format, it makes things up, and users try to trick it. You will see each failure happen in runnable code and learn which phase of the course fixes it.

Every example runs on this page in Python, Java and JavaScript, and prints the same output in all three. Examples use small scripted fake models so they are free and repeatable; the real-model labs let you send requests to a free hosted model and see the real response, token counts and limits. The phase ends with your first playbook: the checklist and starter client for any LLM feature.

0/7 · 0%
7 topics ~2 h 15 code blocks & diagrams
Start with the first topic
1
0.1

What a Large Language Model Is

A large language model (LLM) is a program that reads text and predicts what text should come next, one small piece at a time. Chat assistants are an ordinary app wrapped around that one ability.

20 min 2 code practice

2
0.2

Next-Token Prediction, Built by Hand

You build a tiny language model from word counts, turn the counts into probabilities, and generate text one token at a time, first by always picking the top word (it loops) and then by sampling (it varies). Real LLMs run the same loop with a far better predictor.

21 min 2 code practice

3
0.3

When to Use an LLM (and When Not To)

Use plain code for anything exact (sums, dates, lookups, rules) and an LLM for fuzzy language work. The strongest designs combine both: code handles what it can, the model handles the rest, and code checks the model.

19 min 2 code practice

4
0.4

Why "Just Call the API" Fails in Production

A demo calls the model once and works. A product makes thousands of calls and meets rate limits, timeouts, broken formats, wrong answers, attacks and model updates. Each failure has a known engineering fix, and this course teaches them in order.

21 min 2 code practice

5
0.5

What an LLM Engineer Does

An LLM engineer builds the system around the model: input checks, retrieval of the right data, the prompt, the model call, output validation, tools, evaluation, security and monitoring. This topic maps that stack to the phases of the course.

18 min 1 code practice

6
0.6

How This Course Runs Code

Every key example runs on the page in Python, Java and JavaScript and prints the same output, using small scripted fake models so runs are free and repeatable. Real-model labs send your own requests to a free hosted model and show the real reply, tokens, timing and limits.

19 min 3 code practice

7
0.7

Playbook: Your First LLM Feature

Two recipes to start every LLM feature: a fake model behind the same interface as the real one, so the feature is testable from day one, and a meter that counts tokens, estimates cost and refuses calls over budget. Plus the checklist to run before you ship.

21 min 3 code practice

Practise this phase

Hands-on exercises: write the code, run it, and press Check to have it tested automatically.

1

Count the next words

Write next_counts(text, word) that returns a dictionary: for every place word appears in text (split on spaces), count the word right after it.

Beginner
2

Generate greedily, with a stop rule

table maps a word to a dictionary of next-word counts. Write generate(table, start, max_tokens) that returns a list of words starting with start.

Beginner
3

Check the model's total

A model writes order confirmations like "Your total is Rs 1,159.00.". Write claimed_total(reply) that returns the amount after "Rs " as a float (remove the thousands commas and a final full stop). Return None if the reply has no "Rs ".

Beginner
4

One retry, only for the right errors

Write call_with_one_retry(call). call is a function with no arguments that returns text or raises an error.

Beginner
5

Estimate the cost of a day's calls

Write cost_dollars(usages, price_in, price_out). usages is a list of {"input": n, "output": n} token counts; the prices are dollars per million tokens. Return the total cost rounded to 6 decimals.

Beginner