John Michailidis

With a solid understanding of Generative AI, it’s time to explore how to effectively interact with and optimize AI assistants—without writing a single line of code. In this hands-on session, participants will engage with a custom-built Generative AI assistant powered by Llama, Llama, and DeepSeek-R1. Through guided, no-code exercises, attendees will experience firsthand how to refine AI outputs and maximize performance.

We’ll begin by experimenting with prompt engineering, where participants will explore different ways to structure inputs to guide AI responses. By adjusting tone, specificity, and constraints, they will see how small modifications can significantly improve relevance, coherence, and accuracy.

Next, we will dive into response optimization techniques, demonstrating how settings like temperature, response length, and system instructions impact AI-generated content. Attendees will practice adjusting these parameters through interactive exercises, gaining intuition on how to tailor AI behavior for different use cases.

Finally, we’ll discuss efficiency and best practices, showing how to make interactions smoother, reduce hallucinations, and achieve more reliable outputs. By the end of this session, participants will have a hands-on understanding of how to effectively work with GenAI models, empowering them to leverage AI assistants like a pro.

Generative AI is reshaping industries by enabling powerful applications in content generation, automation, and decision-making. However, to truly unlock its potential, one must understand not just how it works, but also how to optimize it for efficiency and accuracy. This session provides a structured introduction to Generative AI, making it accessible for novice students and AI enthusiasts.

We will start with the fundamental principles of Generative AI, explaining how models like Llama and DeepSeek-R1 generate human-like responses. Participants will gain insights into key architectural components, including transformers, attention mechanisms, and tokenization, as well as the differences between training and inference processes.

Moving beyond the basics, we will explore essential optimization techniques used to enhance model performance. This includes strategies for improving response accuracy, reducing latency, and balancing computational efficiency. Attendees will also learn about the latest advancements in fine-tuning methods, parameter-efficient training, and the trade-offs between different deployment strategies.

By the end of this session, participants will have a strong conceptual understanding of Generative AI, equipping them with the knowledge to build and optimize AI applications in the next hands-on session.