Large Language Models: From Transformer Basics to Agentic AI

This live online course introduces participants to the dynamic world of Large Language Models (LLMs) through a hands-on and practical lens.

The course begins with foundational topics—including text preprocessing, neural networks, and transformer architectures—and quickly moves into applied use cases. Participants will explore how LLMs power real-world applications, from AI-driven customer service to content generation, and beyond.

Participants will learn in this live (synchronous) online course through 19 hours of interactive, expert-led lectures; hands-on coding tutorials; and conversations with industry contributors. Learners will gain direct experience using and modifying models via popular APIs—including Hugging Face and OpenAI—and dive into advanced techniques like Retrieval-Augmented Generation (RAG), vector databases, agentic workflows, and fine-tuning with LoRA.

In addition, pre-program materials will include approximately 4-6 hours of pre-recorded content to help you prepare and get the most out of the program. These materials will be available several weeks before the live program sessions begin and can be completed at your own pace. Pre-program content includes lecture-style modules introducing language models and transformer architectures, along with hands-on tutorials on preprocessing text data (e.g., tokenization, lemmatization, stemming) and using pre-trained models with Hugging Face. Reviewing this pre-program material is highly recommended for all participants. In addition, pre-program materials will also offer an optional refresher on basic Python, linear algebra, and introductory statistics; these materials are designed for those who wish to review foundational concepts that may be helpful during the program.

At the end of the program, there will be an optional, friendly hackathon where you’ll put your new LLM skills to the test in a realistic, hands-on challenge. You will be provided with a starter codebase and a problem statement—then it’s up to you to compete in building the most effective solution. Prizes will be awarded for “Best in Show,” “Most Innovative,” and “Most Elegant” implementations. Whether you’re looking to deepen your learning or just have fun with fellow participants, this optional hackathon is the perfect way to wrap up your program experience.

By the end of the course, participants will be equipped to craft effective prompts, apply advanced LLM techniques, and integrate models into real-world workflows—whether in research, industry, or creative domains.

They’ll also become part of a growing community of applied AI practitioners and receive a certificate of participation from the Harvard John A. Paulson School of Engineering and Applied Sciences.