Artificial Intelligence

AI can be approached through how models learn and compute, or through integrating models into systems that complete tasks. Model selection, agents, retrieval and classifier training share foundations without requiring one compulsory course order.

Choose a Topic by Your Question

Topic	Question
Model and product selection	Which model and interface fit the task?
Local deployment and inference	Will it fit, and how fast will it run?
Agent engineering	How does a model become a controlled tool-using system?
Retrieval, context and memory	Which evidence should reach the model, and what should persist?
AI coding and learning	How do these systems fit coding and learning?
Machine learning and training	What changes during training, and why does it generalize?
Neural and generative models	How do architectures represent inputs and produce outputs?
Data and evaluation	What makes a result trustworthy for its intended use?

Three Reading Paths

Put AI to work. Start with model and API selection to establish capabilities and costs, then organize tasks with the repository workflow. When evidence is missing or excessive, read the retrieval pipeline and context engineering. Use task evaluation to establish a benefit beyond a successful demo.

Understand the model. Begin with learning objectives, linear regression and MLPs. Follow text representations, attention and Transformers into training stages and generation.

Build a local agent. Estimate resources and choose a runtime. Define tool contracts, the execution loop and recovery. Add memory when information must persist across tasks; compare quality, latency and failures on the same work.

Connect Concepts, Implementations and Results

An architecture article should explain inputs, outputs and training objectives. A tool guide should enable a small task and make failures recognizable. A comparison should expose the conditions behind its conclusion. These kinds of notes refer to each other without substituting for each other.

Use the mathematics map for prerequisites and computing map for programming and systems. Look up unfamiliar terms in the data science vocabulary, then return to your question.

My reading · Opened notes are remembered in this browser.