March 12, 2026 | University of Ioannina, Arta Branch
This intensive 2-hour lecture equips university students with both foundational knowledge and advanced expertise in Generative AI development using TypeScript. Blending essential theory with hands-on coding demonstrations, the session explores core principles and state-of-the-art techniques powering today’s production AI systems. Students will gain practical skills and deep insights to confidently build and innovate with Generative AI in real-world scenarios.
CRASH COURSE
2 HOURS
IN PERSON INSTRUCTOR LED TRAINING
HANDS ON FRAMEWORKS AND USE CASES
This session establishes the technical groundwork for building real-world AI applications with modern development practices.
We begin with TypeScript essentials for AI development, focusing on the patterns and language features that matter most when working with APIs, structured outputs, async workflows, and strongly typed AI responses. Participants will understand how to design clean, scalable AI integrations using type safety as a competitive advantage—not an afterthought.
Next, we explore how Generative AI and Large Language Models (LLMs) actually work. Instead of abstract theory, this section explains the mechanics developers need to understand: tokens, context windows, temperature, embeddings, inference, and the probabilistic nature of text generation. By the end, attendees will clearly grasp what happens between prompt and response—and why outputs behave the way they do.
The session then moves into a live code demonstration, showing how to integrate an LLM into an application with real-time streaming responses. Participants will see how to:
Finally, we cover token management and cost estimation, a critical but often overlooked aspect of production AI systems. This includes:
By the end of this 50-minute foundation block, participants will understand not just how to call an LLM—but how to build production-ready AI features responsibly, efficiently, and at scale.
This session moves beyond simple prompt-response patterns into structured reasoning systems and autonomous AI architectures designed for real-world deployment.
We begin with advanced prompt engineering techniques, introducing practical reasoning frameworks that significantly improve output quality and reliability:
Rather than treating prompting as trial-and-error, this section frames it as system design, where reasoning strategies are deliberately engineered to produce predictable outcomes.
Next, we explore AI agent architecture and design patterns. Participants will learn how modern agents are structured, including:
This provides a clear mental model for building agents that move beyond chat interfaces into goal-driven systems.
The session then shifts to a live code demonstration, implementing a ReAct-style agent with tool integration. Attendees will see:
Finally, we address production considerations, focusing on what separates demos from deployable systems:
By the end of this session, participants will understand how to design, implement, and deploy intelligent AI agents that are structured, controllable, and production-ready—not just impressive prototypes.