Hero guide: Generative AI
This guide highlights sessions at re:Invent 2024 to help you stay on top of everything related to generative AI, from the basics of understanding and effectively utilizing generative AI technologies to real-world use cases, Amazon Q, and beyond.
Rustem Feyzkhanov
AWS Machine Learning Hero
Generative AI continues to transform industries, making it crucial for businesses to stay updated on the latest advancements and best practices. Deploying generative AI in production presents unique challenges due to the distinct architecture it requires compared to traditional MLOps. This guide showcases the top re:Invent 2024 sessions to help you navigate the rapidly evolving landscape of generative models.
The sessions I’ve highlighted are available in person and on the re:Invent livestream and are organized into the following themes:
Basics
These sessions lay the groundwork for understanding and effectively utilizing generative AI technologies. They cover foundational concepts, focusing on prompt engineering and key decision-making factors for leaders.
Use cases
An important part of designing great architecture for LLM is understanding success stories and how businesses across various industries are harnessing this technology to drive innovation and efficiency.
Amazon Q
Amazon Q is a generative AI–powered assistant designed to enhance productivity across various business functions. Sessions in this category explore how Amazon Q is revolutionizing business intelligence, employee productivity, and developer efficiency.
Data for generative AI
Data is the foundation of any AI model, especially generative AI. High-quality, well-structured data directly influences the accuracy, efficiency, and relevance of AI applications. These sessions focus on strategies for building a robust data foundation, handling data governance, and optimizing data architectures, which are essential for customizing and deploying AI models effectively.
LLMOps
These sessions address the operational challenges and best practices for deploying and maintaining LLMs in production environments. They focus on the infrastructure, scalability, and cost-effectiveness required to successfully integrate LLMs into business workflows.
Breakout session
TNC101 | Introduction to prompt engineering
In this session, you learn how to create and optimize prompts for a variety of generative AI models. First, this session covers the basics of foundation models (FMs), including a subset of FMs called large language models (LLMs). Then, the session covers the fundamental concepts of prompt engineering, such as the different elements of a prompt and some general best practices for using prompts effectively. Finally, the session provides information about basic prompt techniques, including zero-shot, few-shot, and chain-of-thought prompting.
TNC102 | Generative AI for decision-makers
Over the course of 60 minutes, this session for leaders covers the generative AI factors that help you make clear decisions on projects that incorporate AI. Come learn about the basics of generative AI terminology and the potential benefits and risks of using generative AI. Also, learn the steps for planning a generative AI project and the key considerations for building a generative AI–ready organization.
AIM306 | Tool use & agents at the frontier: Advanced techniques for LLM actions
Base LLMs can read and write but aren't capable of acting on their own. Tool use and agents allow models to connect to your APIs and other real-world systems to turn their knowledge into action. Discover how to use advanced prompt engineering and clever system design to craft powerful automations using Anthropic’s Claude models in Amazon Bedrock, learning specialized tool syntax and optimizing solutions through strategic planning. Also, learn how frameworks like Amazon Bedrock Agents are pushing the boundaries of AI capabilities. Join us to explore the forefront of AI-driven automation and elevate your LLM skills.
DEV301 | The art of transforming foundation models into domain experts
Standard foundation models have general knowledge on most topics, but little specific domain knowledge. Learn how to customize them to make them experts in your field. In this session, learn how to perform fine-tuning, use the Retrieval Augmented Generation technique, or use transactional agents through interactive demonstrations. Discover how to evaluate the pros and cons of each approach to choose the solution that best suits your needs.
NTA301 | Automating scalable order processing using generative AI
AffinityX, a leading advertising agency processing over 3 million purchase orders annually for promotional goods, encountered scalability challenges. In this session, hear how they leveraged generative AI on AWS and Cloudwick’s expertise to develop a solution powered by Amazon Bedrock, Anthropic Claude 3, AWS Lambda, and other serverless services. Discover how this innovative approach automated the entire order workflow, seamlessly handling the high volume of orders from submission to fulfillment without manual intervention. Gain insights into how AffinityX achieved cost savings, scalability, and the ability to dynamically adapt to demand by embracing generative AI on AWS for order processing automation.
BIZ201 | How the U.S. Army uses AWS Wickr to secure mission-critical comms
Seamless and secure collaboration is central to operational success. Join this session to discover how the U.S. Army uses AWS Wickr to protect critical communications between various units, mission partners, and citizens. You’ll learn how your organization can deploy the end-to-end encrypted communication capabilities, data residency controls, and administrative features AWS Wickr provides to protect messages and files, manage users and policies, and retain conversations in a data store of your choice to meet regulatory needs.
BSI101 | Reimagine business intelligence with generative AI
In this session, get an overview of the generative AI capabilities of Amazon Q in QuickSight. Learn how analysts can build interactive dashboards rapidly, and discover how business users can use natural language to instantly create documents and presentations explaining data and extract insights beyond what’s available in dashboards with data Q&A and executive summaries. Hear from Availity on how 1.5 million active users are leveraging Amazon QuickSight to distill insights from dashboards instantly, and learn how they are using Amazon Q internally to increase efficiency across their business.
AIM213 | Boost employee productivity with Amazon Q Business
In this session, explore how to use Amazon Q Business, a generative AI–powered assistant, to remove the tasks that slow your team down. Quickly customize Amazon Q Business to answer questions, provide summaries, generate content, and complete tasks based on data and information in your enterprise systems. Learn how Life360 uses Amazon Q Business to accelerate their 300 engineers to harness information sprawl and move faster. With security and privacy as central considerations, Amazon Q Business respects existing user identities, roles, and permissions, and allows administers to restrict sensitive topics, block keywords, and filter out inappropriate questions and answers.
AIM201 | Maximize business impact with Amazon Q Apps: The Volkswagen AI journey
Discover how Volkswagen harnesses generative AI for optimized job matching and career growth with Amazon Q. Learn from the AWS Product Management team about the benefits of Amazon Q Business and the latest innovations in Amazon Q Apps. Then, explore how Volkswagen used these tools to streamline a job role mapping project, saving thousands of hours. Mario Duarte, Senior Director at Volkswagen Group of America, details the journey toward their first Q App that helps Volkswagen’s Human Resources build a learning ecosystem that boosts employee development. Leave the session inspired to bring Q Apps to supercharge your teams’ productivity engines.
DOP204 | GitLab Duo with Amazon Q: AI-driven DevSecOps for your SDLC
In this session, learn how the Amazon Q Developer differentiating agent capabilities for Gitlab Duo transform the developer experience by speeding up a range of tasks that support you as you research how to get started, evaluate system design, build secure and scalable applications, upgrade existing applications, and optimize application performance. Learn firsthand how capabilities for building, testing, reviewing, and transforming applications faster and more easily frees up teams to focus on experimentation and innovation.
STG201 | Build a data foundation to fuel generative AI
Data has long been considered a strategic asset by organizations, but generative AI puts a renewed emphasis on the importance of a data strategy. Whether you are building your own model or customizing a foundation model, your data is a key differentiator for generative AI, so you need a data strategy that supports relevant, high-quality data. In this session, learn about the data tools that can fuel your generative AI strategy and the common data patterns required to transform your generative AI application from a generic tool to a program that truly knows your business and your customer.
DAT319 | A practitioner’s guide to data for generative AI
In this session, gain the skills needed to deploy end-to-end generative AI applications using your most valuable data. While this session focuses on the Retrieval Augmented Generation (RAG) process, the concepts also apply to other methods of customizing generative AI applications. Discover best practice architectures using AWS database services like Amazon Aurora, Amazon OpenSearch Service, or Amazon MemoryDB along with data processing services like AWS Glue and streaming data services like Amazon Kinesis. Learn data lake, governance, and data quality concepts and how Amazon Bedrock Knowledge Bases, Amazon Bedrock Agents, and other features tie solution components together.
DAT405 | Deep dive into Amazon Aurora and its innovations
With an innovative architecture that decouples compute from storage and advanced features like Amazon Aurora Global Database and low-latency read replicas, Aurora reimagines what it means to be a relational database. Aurora is a modern database service offering unparalleled performance and high availability at scale with full open source MySQL and PostgreSQL compatibility. In this session, dive deep into the most exciting new features that Aurora offers including Aurora PostgreSQL Limitless Database, Aurora I/O-Optimized, Aurora zero-ETL integration with Amazon Redshift, and Aurora Serverless v2. Additionally, learn how the addition of the pgvector extension allows for the storage of vector embeddings and support of vector similarity searches for generative AI.
SVS319 | Unlock the power of generative AI with AWS Serverless
Learn to harness the power of AWS Serverless to build robust, cost-effective generative AI applications in this breakout session. Explore using AWS Step Functions to orchestrate complex AI workflows seamlessly. Gain insights through real-world use cases and patterns covering prompt engineering, batch inferencing, Retrieval Augmented Generation (RAG), virtual assistant, and more. Leave equipped with the knowledge and skills to unlock the true potential of secured, highly scalable, high-performance generative AI applications using serverless workflows. Elevate your AI capabilities in this rapidly evolving field.
AIM203 | Migrating to Amazon Bedrock and accelerating gen AI app development
Hear from two AWS customers as they discuss their journey migrating to Amazon Bedrock and finding success in developing generative AI applications on AWS. Leaders from Forcura, a healthcare workflow management company headquartered in Jacksonville, Florida, share how they facilitate continuity of care and improve business performance for providers using automated workflows, collaboration, and analytics SaaS solutions. Additionally, leaders from Cencosud S.A., the largest retail company in Chile and second-largest in LATAM, discuss how they created an assistant to support their grocery customers' digital shopping journeys.
AIM307 | Reduce FM deployment costs and latency with Amazon SageMaker
Organizations need robust, scalable, and cost-effective solutions to deploy and serve foundation models (FMs). This session explores how to use Amazon SageMaker to deploy FMs to make predictions at the best price performance for any use case. Get a detailed overview of deployment strategies to support large-scale generative AI inferencing, and learn how to architect solutions that optimize performance and cost.
AIM303 | Customize FMs with advanced techniques using Amazon SageMaker
Amazon SageMaker allows data scientists and ML engineers to accelerate their generative AI journeys by deeply customizing publicly available foundation models (FMs) and deploying them into production applications. The journey begins with Amazon SageMaker JumpStart, an ML hub that provides access to hundreds of publicly available FMs, such as Llama 3, Falcon, and Mistral. Join this session to learn how you can evaluate FMs, select an FM, customize it with advanced techniques, and deploy it—all while implementing AI responsibility, simplifying access control, and enhancing transparency.
Closing comments
This guide should help you navigate the key generative AI sessions at re:Invent 2024. Whether you're a decision-maker, developer, or industry professional, these sessions provide valuable insights into how to effectively use generative AI in your organization.