Topics
What we’re making next. Every topic here becomes a video, a page of explanation, and runnable code in this repo. Each page already explains the idea, so you can start learning before the video is out.
Have an idea? Suggest a topic.
Machine learning
| Topic | Hook | Status |
|---|---|---|
| What is machine learning | Watch the video and run the code | Published |
| Gradient descent | Watch the video and run the code | Published |
| K-means clustering | Watch the video and run the code | Published |
| Linear regression, no libraries | Watch the video and run the code | Published |
| A neural network from scratch | One straight line can’t solve this four-point puzzle. Two layers can. | Lesson ready |
| How a decision tree thinks | This model plays 20 questions with your data, and you can read every answer. | Lesson ready |
| Overfitting: why a perfect score is a red flag | This model scored 100%. That’s the problem. | Lesson ready |
| How to choose an ML algorithm in production | There is no best algorithm. Here’s how teams actually pick one. | Lesson ready |
AI engineering
| Topic | Hook | Status |
|---|---|---|
| The AI agent loop | Watch the video and run the code | Published |
| RAG from scratch | Watch the video and run the code | Published |
| Tokenization: why AI can’t count letters | Ask an AI how many r’s are in ‘strawberry’. Here’s why it struggles. | Planned |
| Temperature: how an AI picks the next word | Same prompt, different answer every time. One number controls it. | Planned |
| Embeddings and vector search | Your search found ‘car’ when you typed ‘vehicle’. Here’s how. | Planned |
| Tool calling: how an AI runs your code | The model doesn’t run your function. It writes a request, and your code decides. | Planned |
Methodologies
| Topic | Hook | Status |
|---|---|---|
| Ontologies for AI agents | Watch the video and run the code | Published |
| Evals and loop engineering | Watch the video and run the code | Published |
| LLM as a judge | Grading AI with AI works, until it doesn’t. | Planned |
| Prompt injection | One sentence hidden in a web page can take over your AI agent. | Planned |
Data engineering
| Topic | Hook | Status |
|---|---|---|
| API to Parquet | Watch the video and run the code | Published |
| Incremental loads and change data capture | Your pipeline reloads 10 million rows to pick up 200 changes. | Planned |
| Idempotent pipelines | Your job failed halfway, you reran it, and now revenue is doubled. | Planned |
| ETL vs ELT | Same three letters, different order, and it changes your whole data platform. | Lesson ready |
| PySpark: data too big for one machine | Your laptop chokes on a billion rows. Spark splits the job across a cluster. | Lesson ready |
| Polars: the fast DataFrame library | Same question, same data. One library reads the whole file, the other reads only what it needs. | Lesson ready |
| Documents to data: OCR, Parquet and a vector database | A folder of scanned invoices becomes a table you can query and a knowledge base an AI can search. | Lesson ready |
Data architecture
| Topic | Hook | Status |
|---|---|---|
| Bronze, silver, gold: the medallion architecture | Raw data in, trusted data out, in three layers. | Planned |
| Star schemas: facts and dimensions | Why the analytics team keeps talking about facts and dimensions. | Planned |
Exploratory data analysis
| Topic | Hook | Status |
|---|---|---|
| Simpson’s paradox | Treatment A wins in every group, and loses overall. | Planned |
| When the average lies | Ten people in a bar. A billionaire walks in. Now the average person is a millionaire. | Planned |
Software engineering
| Topic | Hook | Status |
|---|---|---|
| Big O, by experiment | Fast on 100 rows. Dead on a million. Here’s why. | Planned |
| How Git actually stores your code | Git is a key-value store wearing a version control costume. | Planned |
Backend
| Topic | Hook | Status |
|---|---|---|
| Rate limiting with a token bucket | How an API tells you to slow down. | Planned |
| Caching and the hardest problem in computer science | Your app got 50 times faster. Then it showed yesterday’s prices. | Planned |
Cloud
| Topic | Hook | Status |
|---|---|---|
| Docker images, layer by layer | Why your Docker image is 2 GB. | Planned |
| What Terraform state actually is | Terraform remembers what it built. Lose that memory and it builds everything again. | Planned |
| Kubernetes and the reconciliation loop | Delete a container and watch it come back. | Planned |
| How AWS decides allow or deny | Your Lambda has permission, and AWS still says Access Denied. | Planned |
Build Projects
| Topic | Hook | Status |
|---|---|---|
| A production RAG assistant, end to end | From a folder of PDFs to an assistant your team actually trusts. | Planned |
| A data platform on AWS with Terraform | From raw events to a live dashboard, all defined in code. | Planned |