SERVICES
│ Separate-track curriculum
ChatGPT API, embeddings, RAG
A technical-skill course that, with the final goal of "being able to build a system that combines internal information with AI (RAG) yourself," builds up step by step from the basics of the API to embeddings and Transformers.
─ ARTICLE COVER
System Design Skills ／ Separate-track curriculum
§I ─ POSITION
A technical-skill course in AI product development, independent of the DX-project curriculum. Learn while implementing in Python.
§II ─ TAKEAWAYS
Call the ChatGPT API and embed it into operations
Implement the "search" part of RAG with embeddings and semantic search
Understand the internal workings of the Transformer (attention) hands-on
§III ─ EXCERPT
¶ 01
A structure of 3 sessions plus a supplement. Build up while implementing.
| Session | Theme | What you do |
|---|---|---|
| Session 0 | Introduction | An overview of the whole curriculum and the big picture of RAG |
| Session 1 | Running the ChatGPT API | Python setup / API calls / basic parameters |
| Session 2 | The world of embeddings | Vectorizing text / semantic search / implementing the search part of RAG |
| Supplement | How the Transformer works | Experiencing the attention mechanism |
¶ 02
You can take this if you have the basics of Python (if statements / for loops / functions / classes), installing libraries, and basic terminal operations. No experience in web development or machine learning required.
¶ 03
RAG (Retrieval-Augmented Generation) is the structure of searching internal documents → including the relevant passages in the LLM's prompt to generate an answer. In Session 2 you implement embeddings + vector search to build the search part. The generation part is an application of the ChatGPT API calls from Session 1.
¶ 04
If you just implement RAG, the Transformer internals aren't needed. But if you can't answer the field's questions — "what is attention?", "where does the context-length limit come from?" — you'll get stuck in AI-product-development decisions. In the supplementary session, you feel out just the key points while running a verification script.
§∞ ─ DOWNLOAD
─ EXERCISE
Each session has assignments and an instructor's sample answer. The Transformer supplement comes with an experiment-report template and a verification script (transformer_demo.py).
Lecture slides
Every slide of the course
Cheat sheet
Quick reference for the field
Exercises
With answers and grading
Advanced / test
Hands-on plus rubric
§IV ─ RELATED