Let's Code LLM Chatbots and Agents from Scratch

You'll create LLM-powered apps from scratch without needing to rely on frameworks that abstract away the important details.

You'll understand how LLMs work under the hood, and what they can and cannot do.

Understanding the essential nature of LLMs as a statistical next-token predictor is the key for utilizing them effectively.

LLMs are unpredictable by nature, but you'll know how to steer them into doing what you want and achieving your goals.

You'll use prompt and context engineering to tame the nondeterministic LLM and reduce hallucinations.

You'll build chatbots that converse accurately about your organization's data and help advise users appropriately.

You'll create your own search engine that use vector databases to power semantic search.

Instead of simply hoping that your newest updates make things better, you'll use evals to consistently monitor your app's quality.

You'll learn how to perform error analysis, annotate traces, and even automate your evals.

You'll equip LLMs with tools that can do more than generate text - they'll trigger real code functions

Your agents will write code, deploy websites, and even produce podcasts.

You'll gain finer control of your agent by assembling agentic workflows, thereby reducing the ways in which the agent can go rogue.