Absolute Path

AI developer

Build apps that use language models, and check that they work.

An AI developer adds language models to real software. You learn Python, call a model from code, write prompts, let a model search your own documents, and measure whether the app gives good answers.

  • Building now
  • 3 of 6 courses ready
  • About 21 hours in total
  • Free

Not started

Start the path

Your path, step by step

Some courses are ready. The others are being written now. A step with no link is a course that is planned. Its time is an estimate.

  1. Python

    Up next

    Learn a clear, friendly language used almost everywhere.

    • Beginner
    • 10 lessons
    • 1 h 54 min

    Not started

    Why this step: Most AI tools and libraries use Python.

  2. Python for data

    Not started

    Read files, work with tables of data, and answer questions with code.

    • Intermediate
    • 15 lessons
    • 4 h 18 min

    Not started

    Why this step: Models work on data. You learn to read, reshape, and check data in Python.

  3. Model APIs

    Not started

    Call a language model from code, and handle its errors, its cost, and its answers.

    • Intermediate
    • 16 lessons
    • 4 h 11 min

    Not started

    Why this step: You call a language model from code, and you handle errors, delay, and cost.

  4. Prompt design

    Coming soon

    Write instructions that give a model the same good answer each time, and test them.

    • About 2 h 30 min

    Why this step: A clear prompt gives the same good answer each time. You learn how to write one and how to test it.

  5. Apps that search your documents

    Coming soon

    Split documents into parts, search them by meaning, and answer questions with sources.

    • About 4 h

    Why this step: A model does not know your documents. You build a search step that gives it the right text, so that it can answer with sources.

  6. Evaluating AI apps

    Coming soon

    Build test sets and scores, and find out when a change makes an app worse.

    • About 4 h

    Why this step: You cannot improve what you cannot measure. You build test sets and scores, and you find out when a change makes an app worse.