Skip to the page
Tecyfy
Daily Bytes app

AI engineering, zero to hero

A free generative AI (GenAI) engineering course, from what a token is to an agent you can ship.

12 modules in order, each one a guide with runnable Python, one exercise that adds to a final project and 3 questions to check yourself. Where Tecyfy already explains an idea, a demo or a calculator, the module links it instead of explaining it again, so each page teaches only what is new at that step.

See the modules Read module 0: Start here: what an AI engineer builds

What an AI engineer does

An AI engineer builds software on top of large language modelsLarge Language ModelA model trained on a huge amount of text to predict the next token. Further training turns it into an assistant that follows instructions.. The model is bought, not trained: you call it through an API, the same way you call a payments or maps service. The work is everything around that call. You decide what goes into the Context WindowContext WindowEverything a model can see at once while it writes an answer, counted in tokens. When the window is full, something has to be left out., which documents to look up first, which tools the model may use, how to tell a good answer from a bad one and what the whole thing costs a month.

That makes it closer to software engineering than to data science. You will write ordinary code, tests and logs. What is new is that one part of the program answers differently each time and can be confidently wrong, so a large share of the job is measuring it: an eval set that runs on every change, and traces that show why an answer went wrong.

Generative AI, or GenAI, is the wider name for models that write text, code or pictures. This course covers the text and code side, which is where most GenAI engineering jobs are, and uses the Claude API for its examples. Every idea carries over to OpenAI, Gemini and open models, and each module says where a provider does something differently.

The modules

Each module says what you can do at the end and links what Tecyfy already has on it. Mark a module done at the end of its guide, and this map keeps your place in this browser.

Where do you start?

Choose what fits you, and the module to start with is marked below.

Foundations

What the job is, and how a model turns your words into an answer.

  1. Module 0: Start here: what an AI engineer builds

    You can: Say what the job is, set up Python, an API key and a coding agent, and run your first call.

    Read the guide

    With it on Tecyfy

  2. Module 1: How a model writes an answer

    You can: Explain tokens, the context window and sampling, why a model is confidently wrong, and which of these you set in code.

    Read the guide

    With it on Tecyfy

Build with a model

Calls, prompts and search, the parts every GenAI app is made of.

  1. Module 2: Your first real model call

    You can: Call a model with a system prompt, stream the answer, get JSON that fits a schema, handle errors and rate limits, and read what each call cost.

    Read the guide

    With it on Tecyfy

  2. Module 3: Prompts and context

    You can: Keep prompts as versioned templates with test cases, and decide what goes in the context and what stays out.

    Read the guide

    With it on Tecyfy

Build the app

Answers grounded in your documents, tools the model can call, and the evals that tell you it works.

  1. Module 5: RAG that answers correctly

    You can: Build retrieval-augmented generation that cites its sources, and measure whether retrieval found the right text.

    Read the guide

    With it on Tecyfy

  2. Module 6: Tools and agents

    You can: Give a model tools, write an MCP server, run the agent loop with limits, and choose between a workflow and an agent.

    Read the guide

    With it on Tecyfy

  3. Module 7: Evals: knowing it works

    You can: Build an eval set, grade with code, a model and people, tell a real change from noise, and run evals on every change.

    Read the guide

    With it on Tecyfy

Ship it

Safety, cost and speed in production, open models, and a project that puts it all together.

  1. Module 8: Safety and security

    You can: Limit what an injected or wrong agent can do, keep secrets and personal data out, and put a person before actions that cannot be undone.

    Read the guide

    With it on Tecyfy

  2. Module 9: In production: cost, speed and reliability

    You can: Budget a feature, cut its cost and latency, add fallbacks, and watch it once it is live.

    Read the guide

    With it on Tecyfy

  3. Module 10: Open models: run and adapt your own

    You can: Size a GPU for an open model, serve it, and know when fine-tuning beats prompting and RAG.

    Read the guide

    With it on Tecyfy

  4. Module 11: Capstone: a support assistant, end to end

    You can: Build an assistant that answers from a company's documents with citations, looks up an order with a tool, asks a person before a refund, refuses an injected instruction and passes its own eval set within a cost budget.

    Read the guide

    With it on Tecyfy

After the course

Test what you know, then practise the questions AI engineer interviews ask.

Questions

What is AI engineering?
AI engineering, also called GenAI engineering, is building software products on top of large language models that you call through an API. The engineer chooses the model, writes the prompts and the context, connects documents and tools, measures the answers with evals and keeps the cost and speed in check. Training a model from scratch is a different job.
Is this AI engineering course free?
Yes. Every module, exercise and question is free, with no account and no sign-up. Running the code calls a model API, which its maker bills by the token; the exercises are small, and module 2 shows how to read what each call cost.
What do I need to know before I start?
Enough Python to write a function, install a package and run a script. You do not need machine learning, statistics or a GPU. Module 0 sets up Python, an API key and a coding agent, and module 1 explains how a model works in plain terms.
How long does the course take?
About 2 hours a module, reading the guide and doing its exercise, so about 24 hours in all. You can start at any module that fits what you know; the map at the top of this page suggests where.
What is the difference between an AI engineer and a machine learning engineer?
A machine learning engineer trains and serves models, so the work is data, training runs and model quality. An AI engineer builds products on models someone else trained, so the work is prompts, retrieval, tools, evals and cost. Module 10 covers where the 2 meet, when you run or fine-tune an open model yourself.
Why does the course use the Claude API?
Each example needs one real API to run, and the Claude API covers everything the course teaches, from streaming and JSON output to tools and prompt caching. The ideas carry over, and where OpenAI or Gemini do something differently the module says how.

Sources

Where each fact comes from

  1. 1Messages API
  2. 2Building effective agents
  3. 3Define success criteria and build evaluations