Developing Advanced Generative AI Applications 

Duration: 3 Days / 24 hrs

Introduction 

This 3-day, hands-on course focuses on building advanced Generative AI applications that go beyond basic prompting. Participants learn how to fine-tune models efficiently, design memory-aware chatbots, work with multi-modal inputs, and build agentic, autonomous AI systems using modern frameworks. 

Objectives 

By the end of this course, participants will be able to: 

  • Apply parameter-efficient fine-tuning techniques like LoRA and IA3 
  • Build GenAI applications with memory, persistence, and long-context handling 
  • Implement tool calling, prompt chaining, and advanced output strategies 
  • Design agentic and autonomous AI systems using industry-grade frameworks 

Key Takeaways 

Participants will leave with: 

  • Hands-on experience building production-grade GenAI systems 
  • Practical understanding of fine-tuning, multi-modal prompting, and orchestration 
  • Confidence to design scalable, secure, and compliant AI applications 

Training Methodology (Learning by Doing) 

  • Hands-on labs across all modules—no slide-only sessions 
  • Incremental builds from fine-tuning to agentic workflows 
  • Real-world scenarios including risk, cost, and compliance considerations 
  • Build, test, refine with continuous evaluation and debugging 

Course Outline 

Parameter-Efficient Finetuning Techniques 

  • Comparison with traditional finetuning 
  • Prompt-based methods 
  • Low-Rank Adaptation (LoRA), IA3, and others 
  • Performance evaluation 

Chatbots with Memory and Persistence 

  • Persisting messages 
  • Handling long conversations 
  • Cross-session memory 
  • Fact extraction and privacy considerations 

Multi-modal Prompting and Interaction 

  • Processing non-textual inputs 
  • Prompting with images and text 
  • Stable Diffusion overview 
  • Cost optimization 
  • Use cases 

Streaming, Ensembling, and Output Techniques 

  • Streaming LLM outputs 
  • Self-consistency and post-processing 
  • Mixture methods and self-refinement 

Function and Tool Calling 

  • Binding functions to LLMs 
  • Handling responses and errors 
  • Security and retries 
  • LangChain Toolkits 

Advanced Use Cases 

  • Embedded agents 
  • Agent-based modeling 
  • Code generation and cybersecurity 
  • Risk management and EU AI Act 

Prompt Chaining and Directed Graphs 

  • Predetermined and generated paths 
  • Context passing 
  • Conditional routing and debugging 

Agentic AI and Autonomous Systems 

  • Definitions and characteristics 
  • Comparison with traditional AI 
  • Real-world applications 
  • Architectures: Reactive, Deliberative, Hybrid 
  • Frameworks: AutoGen, LangGraph, CrewAI