AWS GenAI Bootcamp | Day 9: Learn FMOps/LLMOps using Bedrock | Hosted by AWS UG Dehradun

AWS GenAI Bootcamp | Day 9: Learn FMOps/LLMOps using Bedrock | Hosted by AWS UG Dehradun

Ladies and gentlemen, a very good evening to one and all present here. I'm Unati Agarwal, a co-member of the AWS User Group Dehradun (AWS UG D), and I'm thrilled to welcome you to day nine of our AWS Generative AI Bootcamp.

As we have already covered significant ground, it has been an incredible journey. On July 16, we had an enlightening session where we learned about conversational AI and chatbots. Today, we promise to deliver even more value as we delve into the crucial topics of foundational models and large language models (LLMs) using AWS Bedrock. Additionally, we will announce the best party rock apps in today's session, so stay tuned and engaged. Thank you all for joining us on day nine. Your presence here reflects your commitment to learning and exploring cutting-edge technologies in AWS and generative AI.

Today, we are honored to have J Krishnan with us, an AWS Community Machine Learning (ML) Builder, speaker, blogger, mentor, and co-organizer of AWS UG Bangalore. He is an AWS Certified Professional with expertise ranging from generative AI to Azure and GCP. His extensive experience will provide valuable insights into building advanced AI solutions.

We have an exciting agenda: J will begin with insightful discussions on advanced techniques. We will learn how to utilize AWS tools effectively for building a generative AI Q&A chatbot with RAG architecture and deploying foundational models easily with AWS Bedrock. Please stay engaged and participate actively throughout the session to secure a chance to win exciting rewards.

Now, without further ado, I hand over the stage to Mr. J Krishnan. Let's get started!

Introduction by J Krishnan:

Thank you, AWS UG Dehradun, for the introduction. This is the ninth session, and you all have been excitingly running through the fundamental concepts to industry trends in generative AI.

Detailed Agenda and Key Elements:

• Introduction: J Krishnan introduces himself as a seasoned professional with extensive AWS certification and experience. He focuses on sharing knowledge, mentoring, and providing valuable insights into generative AI.

• Overview: Mr. Krishnan shares a comprehensive overview of the journey with large language models (LLMs). He discusses prompting, prompt management, evaluating models, and the FM Ops (Foundational Model Operations) and LLM Ops (Large Language Model Operations) ecosystem.

• Demo Section: The demo covers:

  • Playground: Exploring AWS Bedrock's model playground and performing basic prompts.
  • Prompt Management: Versioning and managing prompts in Bedrock.
  • Model Evaluation: Evaluating models based on selected use cases.
  • Simple Application Building: Using AWS CDK, SAM templates, and API Gateway for a small generative AI application.

• RAG Architecture: Understanding the Retrieval-Augmented Generation (RAG) architecture, including data injection, embedding models, and vector databases for reducing hallucination and grounding models with internal data.

• Agents: Introduction to agents in AWS Bedrock, their application, and orchestration for complex tasks connecting with third-party APIs.

• Fine-tuning the Model: Fine-tuning and continuous pre-training in AWS Bedrock: exploring ways to further refine models based on specific business use cases.

• Real-World Use Case: A comprehensive, end-to-end enterprise-level solution using AWS Bedrock for analyzing financial statements.

• Best Practices and Key Roles: DevOps roles and the significance of community engagement and certifications in AWS.

Keyword

• Keywords: AWS, AWS Bedrock, Generative AI, FMOps, LLMOps, Foundation Models, Large Language Models, RAG Architecture, AI Solutions, Machine Learning, Prompt Management, Model Evaluation, Agents, Fine-tuning, Certifications, Community Engagement.

FAQ

1. What is AWS Bedrock?

   • AWS Bedrock is a managed service by AWS, offering multiple foundational models and capabilities for building and deploying generative AI applications.

2. What is FMOps/LLMOps?

   • FMOps (Foundational Model Operations) and LLMOps (Large Language Model Operations) involve the practices and tools for managing, fine-tuning, deploying, and monitoring foundation and large language models.

3. How do I choose the right foundational model?

   • Identify the use case, test prompts in the AWS playground, evaluate models with AWS’s model evaluation tools, and consider business value against the performance and cost.

4. What is Retrieval-Augmented Generation (RAG)?

   • RAG combines retrieved relevant documents or data with the generated content to produce more accurate and context-aware responses.

5. What are agents in AWS Bedrock?

   • Agents in Bedrock orchestrate and automate tasks by understanding input, selecting appropriate actions, and interacting with APIs or data sources.

6. Why is governance important in generative AI?

   • Governance ensures that generative AI models comply with ethical standards, data privacy, and security guidelines, and helps prevent misuse or unauthorized access.

7. What are the best practices for managing generative AI in an enterprise setting?

   • Implement end-to-end workflows, leverage DevOps principles, ensure data security, continually monitor model performance, and engage cross-functional teams.