The notebook
Notes on software, games, AI, and whatever I’m currently curious about.
From the notebook
Nginx is often introduced simply as a web server, but in modern web applications it commonly plays a broader role. It can sit in front of an application, receive incoming web traffic, handle HTTPS, serve static files, and forward requests to backend services.
The article introduces the core mathematics needed to begin machine learning without overwhelming the reader with advanced theory. It uses a simple student-score prediction example to explain mean, standard deviation, probability, linear relationships, functions, prediction error, vectors, and matrices. The central idea is that AI math becomes much easier when each concept is tied to a practical purpose. The article shows how data is summarized, how relationships are identified, how models turn inputs into outputs, and how prediction errors can be measured using Mean Absolute Error. It also reinforces that libraries such as NumPy handle the calculations, while the reader’s job is to understand what those calculations mean. The article ends by preparing the reader for the next step: training a real machine-learning model.
This article explains how machine-learning systems represent real-world information as numerical data. It introduces NumPy arrays and shows how they differ from regular Python lists, then builds the foundation for understanding scalars, vectors, matrices, dimensions, and shapes. Through practical examples, it shows how students, images, text, and other real-world data can be converted into numerical structures that AI models can process efficiently.
Once you move from basic Python into data science, machine learning, or AI, one of the first things you need to become comfortable with is working with tables of data. Most AI projects do not begin with a clever model or a complicated algorithm.
If you already know how to write a variable, create a function, run a loop, and use an `if` statement, you know enough Python to start moving toward AI.
Artificial intelligence is powerful, but that does not mean every problem should be solved with AI. In many situations, a simple rule, a spreadsheet formula, or basic automation can solve the problem faster, more cheaply, and more reliably. A good AI project begins by asking a more important question than “How can we use AI?” The better question is: “Does this problem actually need AI?”
Artificial intelligence can sometimes look like magic. You give a computer thousands of examples, train a model, and suddenly it can recognize spam emails, detect defects in products, analyze medical images, or answer questions about documents.
Can we estimate what a house will sell for? Can we tell whether a customer is likely to cancel a subscription? Can we identify a defective product before it leaves the factory?
Computers are often described as machines that follow instructions. In traditional programming, a developer writes rules that tell a computer exactly what to do. For example, a programmer might create a rule stating that an email should be marked as spam when it contains certain suspicious words.
Artificial intelligence can sound like a giant robot brain hiding behind a glowing screen. People talk about AI, machine learning, deep learning, neural networks, generative AI, large language models, computer vision, and natural language processing as if they are all the same thing.