GenAI Hands-on Training – Enterprise

Ensure you have the necessary setup and foundational knowledge before starting this course.

A 15-minute session to introduce ourselves, set expectations, and align our learning goals for the course.

Essential session details, connection instructions, and administrative guidelines to ensure a smooth learning experience.

• Basic Principles and Terminology in ML & DL
• Overview of the Model Training Pipeline

• A short break to refresh, recharge, and get ready for the next session.
• Use this time to grab a coffee, stretch, or chat with fellow participants!

• Perceptron, activation functions, loss functions
• Architecture of a simple neural network

• Open Q&A and discussion on key neural network concepts
• Clarify doubts and review practical applications
• Engage in interactive learning with real-world examples

• Key takeaways summarizing the session
• Download course materials and additional resources
• Further reading & references to deepen understanding

• Step-by-step implementation and visualization
• Implementation of a simple NN using Python 

• Fundamental NN architectures: Feedforward, CNNs, RNNs, LSTMs
• Generative & modern AI models: Autoencoders, GANs, Transformers, Diffusion Models
• Deterministic vs. stochastic learning in deep networks

A short break to refresh, recharge, and get ready for the next session.

Use this time to grab a coffee, stretch, or chat with fellow participants!

• OpenAI API vs. Local Deployment (Llama, Mistral, Gemma, etc.)
• Setting up and running a local LLM

• LLM Model Sizes & Training
• Efficiency & Deployment
• LLM Selection & Pricing

• Open Q&A and discussion on key LLMs concepts
• Clarify doubts and review practical applications
• Engage in interactive learning with real-world examples

• Key takeaways summarizing the session
• Download course materials and additional resources
• Further reading & references to deepen understanding

Discuss homework and challenges

This lesson introduces Retrieval-Augmented Generation (RAG) — a technique that enhances language models by allowing them to access and retrieve relevant external knowledge during inference. 

This lesson explains what embeddings are, how they’re generated, and how they power similarity search in vector databases.
We’ll explore different embedding types, how similarity is measured, and why all of this is crucial for Retrieval-Augmented Generation (RAG).

A short break to refresh, recharge, and get ready for the next session.

Use this time to grab a coffee, stretch, or chat with fellow participants!

In this lesson, you'll learn what data chunking is, why it’s essential for working with long documents in RAG and LLM pipelines, and how different chunking strategies — from fixed-size to semantic — affect retrieval quality. You'll also explore when to use each strategy based on your data and use case.

This lesson covers the retrieval phase in Retrieval-Augmented Generation (RAG). You’ll learn what retrieval is, why it’s critical, how it depends on embeddings and chunking strategies, and how it directly affects the relevance, accuracy, and latency of RAG responses.

• Open Q&A and discussion on key LLMs concepts
• Clarify doubts and review practical applications
• Engage in interactive learning with real-world examples

• Key takeaways summarizing the session
• Download course materials and additional resources
• Further reading & references to deepen understanding

Discuss homework and challenges

Final Projects Requirements & Goals discussion

This lesson covers how to collect, clean, and structure data before it's used in a RAG pipeline. We’ll explore different formats, preprocessing methods, enhanced chunking techniques, and how to choose the right vector database. The goal is to maximize retrieval accuracy and efficiency.

A short break to refresh, recharge, and get ready for the next session.

Use this time to grab a coffee, stretch, or chat with fellow participants!

This lesson explores the most effective retrieval strategies beyond basic similarity search. We’ll dive into filtering, hybrid retrieval, multi-vector search, metadata scoring, and how these methods impact RAG performance. Advanced techniques help improve relevance, reduce hallucination, and tailor answers to specific use cases.

This lesson focuses on making RAG systems more intelligent and conversational. We’ll explore how to track user history, detect intent, and auto-augment queries to provide better, more contextually accurate responses — especially in multi-turn conversations or complex search flows.

• Open Q&A and discussion on key RAG concepts
• Clarify doubts and review practical applications
• Engage in interactive learning with real-world examples

• Key takeaways summarizing the session
• Download course materials and additional resources
• Further reading & references to deepen understanding

Discuss homework and challenges

• Single-agent vs. multi-agent systems
• LLMs as reasoning agents

A short break to refresh, recharge, and get ready for the next session.

Use this time to grab a coffee, stretch, or chat with fellow participants!

• Design, implement, and automate agent behaviors with structured workflows for smarter, scalable AI systems

• Open Q&A and discussion on key RAG concepts
• Clarify doubts and review practical applications
• Engage in interactive learning with real-world examples

• Key takeaways summarizing the session
• Download course materials and additional resources
• Further reading & references to deepen understanding

-

• Understand how LLMs use tokenization, embeddings, and transformer architecture.

• Learn core NLP concepts: text representation, syntax vs semantics, and similarity search.

• Explore how prompts function as structured interfaces for LLMs.

• Learn about prompt anatomy, types (zero/few-shot, CoT, ReAct), and token budgeting.

•  

• Apply reusable prompt design patterns like Chain-of-Thought and ReAct.

• Create modular, versioned, and composable prompt templates for scalable use.

A short break to refresh, recharge, and get ready for the next session.

Use this time to grab a coffee, stretch, or chat with fellow participants!

• Use few-shot learning, vector search, and memory to adapt prompts at runtime.

• Design prompts that evolve with user context, session history, or retrieved knowledge.

• Debug, trace, and monitor prompts using LangChain, LangSmith, PromptLayer.
• Build multi-step prompt chains, ReAct agents, and safety-aware LLM pipelines

• Continuously evaluate, A/B test, and optimize prompt performance.

• Implement CI/CD pipelines for prompt versioning, scoring, and feedback-driven refinement.

• Open Q&A and discussion on key prompting concepts
• Clarify doubts and review practical applications
• Engage in interactive learning with real-world examples

• Key takeaways summarizing the session
• Download course materials and additional resources
• Further reading & references to deepen understanding

• Explore LLM-powered tools like Copilot, Sourcegraph, and Custom GPTs

• Map tools to the realms of Code, Backlog, and Documentation

• Understand where GenAI tools fit across the software development lifecycle

A short break to refresh, recharge, and get ready for the next session.

Use this time to grab a coffee, stretch, or chat with fellow participants!

• Generate system diagrams from specs using prompts and LLMs
• Use formats like Mermaid and UML for visual documentation
• Validate and refine outputs to reflect real system architecture

• Open Q&A and discussion on key GenAI tools
• Clarify doubts and review practical applications
• Engage in interactive learning with real-world examples

• Key takeaways summarizing the session
• Download course materials and additional resources
• Further reading & references to deepen understanding

Telegram Bot builder

• Understand how observability reveals the internal behavior of LLM systems through logs, metrics, and traces.

• Learn to track prompts, responses, token usage, and latency for debugging, cost control, and reliability.

• Explore tools like LangSmith, Prometheus, and OpenTelemetry to monitor and evaluate LLM applications in production.

• Open Q&A and discussion on key RL & Observability concepts
• Clarify doubts and review practical applications
• Engage in interactive learning with real-world examples

A short break to refresh, recharge, and get ready for the next session.

Use this time to grab a coffee, stretch, or chat with fellow participants!

• LLMs can return data in structured formats like JSON or YAML for easier parsing and automation.

• Useful in pipelines, agents, and config generation where precision matters.

• Tools like Pydantic, Guardrails, or Outlines help validate outputs.

• LLMs can return structured arguments to trigger named backend functions.

• Enables real-time API calls, tools, and agent actions.

• Safer and more deterministic than parsing free-text commands.

• Key takeaways summarizing the session
• Download course materials and additional resources
• Further reading & references to deepen understanding

Recap & Final Q&A