Mastering Generative AI & Machine Learning

This course contains the use of artificial intelligence. Lectures are narrated by Katie, an AI voice, and AI tools helped produce the scripts, animations, quizzes and the promo video, where Vinay’s footage and voice are AI-generated from his own photos and recordings. The CodeSierra team curated the curriculum, and every notebook was run and tested.

This course goes past the demo. You build something complete: a retention and support assistant for NimbusTel, a fictional telecom company. A support agent types a customer ID.

The assistant looks up the customer, predicts their churn risk (how likely they are to leave) with a machine learning model, searches the company's policy documents and drafts a retention offer that quotes those policies and passes the company's offer rules. The agent reviews it, then approves and saves it. To get there, you learn both halves of modern AI and how to join them.

Predictive AI. You start with Python from scratch, then explore and clean data with pandas, train and evaluate models with scikit-learn, and handle the problems that break real projects: data leakage, imbalanced classes and accuracy scores that hide a useless model. You build a churn model on a public dataset of 7,043 telecom customers.

Deep learning and generative AI. You build neural networks in PyTorch, see how transformers and large language models work, and learn why they hallucinate. You write prompts that hold up, call models from Python through an API, and get reliable structured outputs.

AI applications. You build retrieval-augmented generation (RAG) over documents, give a model tools to call, add guardrails against prompt injection, test your app with evaluation cases, and put a Gradio interface on it. Responsible AI.

You measure bias in your own model, explain its predictions with SHAP, and learn which data should never leave your machine. The capstone ties it all together. You design, build, test and deploy the NimbusTel assistant to Hugging Face Spaces (or a free temporary public link), and finish with a portfolio project you can explain in an interview.

How the course works. 18. 6 hours of video across 10 modules.

Short animated lectures explain each idea; hands-on lectures build notebooks cell by cell with real outputs. Every hands-on lecture ends with a practice exercise, and every module ends with a quiz. Everything runs in Google Colab in your browser.

Modules 6–10 use the OpenAI API (a few US dollars of credit covers the whole course). Lecture 6. 15 shows how to run chat calls on an open model; the search, image and capstone lectures use the OpenAI API.