Course introduction

- [Instructor] Welcome to "Exploring Deterministic LLM Programming." This course teaches you how to explore large language model programming with a flair for determinism, so the idea here is that, instead of just accepting the outcome, you're going to put guardrails on to make your codebase more predictable. If you look at the foundational concepts here, we have performance and optimization. That's the main idea for all of the topics of module one. We also get into warming strategies, Jevons paradox, ELO rating systems, also automation myths, Amdahl's Law and how it could impact also agents, the HAL framework, and then what about the application of ELO for the rating itself? Is it really a one-to-one, or are there some things we should consider? We also talk about advanced techniques in the next module, and this is things like context and control. Some of the topics we cover are the illusion of control, so if you're coding with agents and you let the agents run wild, there's an illusion of control, really, things are degrading. Same with AST and agentic coding. If you're able to use the abstract syntax tree, you can find insights about agents. We also look at the concepts of entropy of code, duplication and deduplication. We look at a technical debt roadmap, a provability score, so how you can make your code mathematically provable. We look at the concept of a deep context, so how can you look at an entire AST, annotate it, and then look at key factors across your codebase in one spot? Also, complexity. Cyclomatic complexity is a huge factor with code quality. We talk about why it's important to have your code have a complexity of 10 or lower, and we also get into the code churn analysis and self-admitted technical debt. Finally, at the very end, we get into a demo of using PMAT. PMAT shows how it can look at complexity and self-admitted technical debt and churn, all these other quantitative measures of code, and then we get into a real repo and walk through how you can use PMAT yourself to make deterministic solutions. All right, we have a lot to cover here. Let's go ahead and get started.