Choosing the right AI/ML and generative AI tools for your use case (Level 200)
AI/ML techniques are important fundamentals for organizations looking to reinvent customer experiences and deliver their objectives. However, applying the right techniques to the use cases is not easy. This session provides guidance on how to apply practical and proven AIML techniques on key use cases for real business impact. We outline existing traditional AI and emerging generative AI suite of services to help you understand when and what services are best suited for the key application requirements or use cases across business functions and industries. We share best practices on how to future-proof your stack and ensure flexibility and control within your organization. This session concludes with how to apply technologies to your use cases, enabling you to conceptualize new opportunities, achieve competitive advantage, and deliver organizational outcomes.

Speaker: Vatsal Shah, Principal Solutions Architect, AWS India
Duration: 30mins

Architecture patterns for building generative AI applications (Level 200)
Do you wish to have guidance on the right tools to build cost-effective and high-performance generative AI applications that are customized for your workload and traffic patterns? In this session, learn the usage patterns and techniques for key use cases such as text generation, summarization, Q&A, chatbot, and image generation to improve productivity and create organizational value. We discuss key considerations when to apply RAG, fine-tuning, or prompt engineering to improve generative AI performance. The session also covers when and how to use advanced prompt engineering, fine-tuning, and RLHF options to improve the results. Find out how to leverage generative AI models with AWS services such as Amazon SageMaker and Amazon SageMaker JumpStart in use cases such as text summarization, simplification, and tone augmentation.

Speaker: Praveen Jayakumar, Head of AI/ML Solutions Architect, AWS India
Duration: 30mins

Cost-optimizing AI/ML workloads on AWS (Level 200)
Many organizations are adopting AI/ML either as a core component or as a supporting workload to achieve high application performance, create delightful user experiences, build sustained competitive advantage, and manage costs effectively. In this session, learn about various ways to cost-optimize your AI/ML workloads. We will delve into ML cost management, highlighting innovative approaches, instance optimizations, lifecycle configurations, and the integration of Amazon SageMaker with Amazon EC2 Spot Instances and Managed Spot Training for Amazon SageMaker and other services. Additionally, we will explore how you can leverage the latest features of Amazon SageMaker to reduce expenses while maximizing the value of your ML workloads.

Speaker: Yudho Ahmad Diponegoro, Senior Solutions Architect, AWS
Duration: 30mins

Select the right large language model for your generative AI use case (Level 200)
There are a large number of large language models (LLMs) out there and choosing the right one is critical because of the high cost associated with deploying generative AI models. Join this session as we explain the tools and considerations when evaluating LLMs. We share how to evaluate LLMs for tasks where the output is fact-based and when the output is creative by nature. With thousands of text generation models out there to choose from and endless prompt engineering possibilities to use them with, learn how you can quickly and reliably identify the best price-performance solution for your use case. Find out how to build a complete picture of model and prompt-template performance on AWS. We showcase how to use automated tools that work alongside human labelers to create scalable but accurate evaluations, enabling you to build high-quality solutions faster and deploy with confidence.

Speaker: Alex Thewsey, Senior AI/ML Solutions Architect, AWS
Duration: 30mins

LLMOps: Lifecycle of a LLM (Level 200)
The rise of large language models (LLMs) surfaced new challenges in development, deployment, and maintenance of these models. While many standard MLOps practices apply, foundation models require additional considerations. Join this session to learn practical tips on adopting LLMs at scale in your organization and maintaining these models in the long run. We explain how you can operationalize generative AI applications using MLOps principles leading to foundation model operations (FMOps).

Speakers: 
Sara van de Moosdijk, Senior AI/ML Partner Solutions Architect, AWS
Vasileios Vonikakis, Senior AI/ML Partner Solutions Architect, AWS

Duration: 30mins

Generative AI platform on AWS (Level 200)
Large language models (LLMs) are growing in popularity but it also surfaced new challenges. Multiple teams need to collaborate and have the right workflows and platforms in place for the deployment and management of generative AI solutions in production. Join this session as we dive into the generative platform capabilities and key considerations for deployment and management of generative AI solutions in production. Learn how to leverage various capabilities at different stages when consuming and iterating on LLMs for different use cases, such as prompt templates management, validation system, feedback system, conversation management, caching and more. The session covers MVS (Minimum Variable Service) and best practices in applying the different LLM models for your use cases. This session also covers the approaches to remove heavy lifting in infrastructure management, enabling your technical teams to focus on core tasks instead.

Speakers: 
Hao Fei Feng, Senior Cloud Architect, AWS Professional Services
Bin Liu, Senior Cloud Architect, AWS Professional Services

Using generative AI responsibly and securely on AWS (Level 200)
Generative AI presents significant opportunities for organizations across multiple industries and it can be a force for good, in areas such as biomedical research and sustainable materials design. But it is important to build generative AI responsibly, securely, and achieve the right balance between innovation and safety. In this session we discuss the key considerations on how to use generative AI responsibly, including fairness, explainability, robustness, privacy, security, governance, and transparency. We dive deep into the operational security features of key AI/ML solutions such as Amazon Bedrock, Amazon CodeWhisperer, and Amazon SageMaker JumpStart. Understand how these solutions provide you with built-in security, privacy, encryption, access control, and compliance, making it easier for you and your organization to build generative AI responsibly into your workflows while maintaining security, as well as adhering to the risk and compliance regulations.

Speaker: Michael Stringer, Principal Security Solutions Architect, AWS
Duration: 30mins

Transform your organization with intelligent document processing (IDP) on AWS (Level 200)
Organizations across several industries often process documents in their daily processes as these documents contain critical information that are required for fast and accurate decision making. However, majority of these documents are still processed manually for extraction of information and insights, which is time-consuming, error-prone, expensive, and difficult to scale. In this session, we explain how to build an intelligent document processing (IDP) solution on AWS to automate information extraction from documents even with different types and formats, enabling your organization to achieve speed and accuracy, without the need for ML skills. Uncover how this solution enables your organization to make quality business decisions faster, freeing your human resources to work on higher value tasks, and reduce overall costs.

Speakers: 
Melwin Pais, Senior Solutions Architect, AWS
Leah Schimer, Associate Solutions Architect, AWS

Duration: 30mins

Build generative AI applications with no code/low code solutions on AWS (Level 200)
In this session, find out how to build an application using natural language with a conversational chat interface to perform tasks such as creating narratives, reports, answering questions, summarizing notes, and providing explanations, without writing a single line of code. We start with demonstrating how to create your own generative AI-based application without writing any code using PartyRock, an Amazon Bedrock Playground. Learn the techniques and capabilities needed to take full advantage of generative AI, including experimenting with various foundation models, building intuition with text-based prompting, and chaining prompts together. We then dive into how you can build and disseminate information to your users quickly, ensuring this information stays entirely within your environment. We also showcase the steps to assemble reports from tabular data in excel or CSV format. This session concludes with guidance on how to easily build, test, compare model outputs and share the findings with relevant teams.

Speaker: Priya Jathar, Solutions Architect, AWS India

Personalize content with generative AI and Amazon Personalize (Level 200)
Organizations are focused on ways to deliver highly personalized user experiences at scale to achieve higher customer engagement, conversion, and revenue while creating meaningful differentiation. In this session, we showcase how you can use Amazon Personalize with generative AI to boost your user engagement and provide highly-optimized customer interactions. Discover how to use Amazon Bedrock LLM foundation model with algorithms from Amazon Personalize to automatically generate thematic connections between recommended content for any interface. We also demonstrate how to build a custom solution with personalized content descriptions that can be integrated into your existing websites, applications, and email marketing systems with simple APIs.

Speakers: 
Tim Wu, Senior AI Specialist Solutions Architect, AWS
Tristan Nguyen, Specialist Solutions Architect, AWS

Duration: 30mins

Bringing the LLM closer to your SQL databases via Agents for Amazon Bedrock (Level 300)
Organizations today still struggle trying to get tangible and measurable value from data. One of the key reasons is because the data are in different formats and structures, and spread out across data warehouses, cloud databases and other systems. Working on data sets to get insights by leveraging multiple solutions in different layers across databases and APIs add further complexity. In this session, we showcase how to access, analyze diverse data sets and generate insights from structured databases. We demonstrate the use of generative AI, analytics and serverless compute to unlock natural language capabilities, enabling new ways to communicate with your existing systems. Discover how to apply to data sources across multiple locations using the same machine learning model without the need to change the computer language in use. We also explain how this simple, pay-as-you-go (PAYG) pricing model provides you and your stakeholders across the organization the ability to access, analyze the diverse data assets to deliver value from your organizational data, without requiring deep technical skillset and resources.

Speakers:
Sam Gordon, Senior Cloud Architect, AWS Professional Services
Ed Fraga, Cloud Architect, AWS Professional Services

Duration: 30mins

Harnessing foundational models for enhanced image creation and search using Amazon Bedrock (Level 200)
Visual content creation is growing in popularity among organizations of all sizes as it enables them to build stronger customer engagement and driving sales. But there are ensuing challenges in effectively managing, searching, and utilizing this vast array of images across various domains. In this session, we discuss the latest advancements in AI-driven image generation and search technologies, with focus on Foundational Models (FMs) on Amazon Bedrock. We explore the ease of integrating, customizing these models as well as underlining the significance of data privacy and security. The session includes a discussion about responsible AI, and approaches to ensure these powerful tools can prevent the generation of harmful or biased content. We also demonstrate how you can build transformational experiences using images in Amazon Bedrock, enabling you to efficiently scale your workflow while conserving valuable time and resources.

Speaker: Suman Debnath, Principal Developer Advocate, Data, AWS
Duration: 30mins

Build impactful insights with no code AI/ML and generative BI (Level 200)
Dashboards are meant to answer questions quickly, enabling all users throughout the organization to improve productivity and provide better decision-making from the data. In this session, learn how generative BI capabilities in Amazon QuickSight enables you to author dashboards using natural language. We demonstrate how business users can easily dive deep into data by simply asking questions. Discover how your users can identify, surface and leverage meaningful insights quickly with Amazon QuickSight with no coding required.

Speaker: Michael Armentano, APJ GTM Lead, Generative BI and Amazon QuickSight, AWS
Duration: 30mins

Extend ML capabilities to relational database-driven applications using AWS no-code/low-code solution (Level 200)
Organizations are looking at ways to improve the data stored in their relational databases and incorporate up-to-the-minute predictions from ML models. However, most of the ML processing are done offline in separate systems, causing delays in receiving ML inferences for use in applications. In addition, developing, installing, and integrating ML models require deep domain knowledge, technical skillset, and appropriate infrastructure. Join this session as we showcase how to extend ML capabilities via digital payment fraud prevention demo using Amazon Aurora ML integration with Amazon SageMaker. We explain how to train models, host endpoints and effectively incorporate real-time model inferences in your applications without any ML training.

Speaker: Darshit Vora, Senior Startup Solutions Architect, AWS India
Duration: 30mins

Improve customer experience with generative AI-powered contact centers (Level 200)
Cloud is changing the way how call centers can accelerate innovation, uncover real-time insights from data quickly, and deliver impactful customer experiences at scale. In this session, discover how to use the new generative AI capabilities built into Amazon Connect to deliver immediate outcomes for your contact center. We dive deep into some of the key features such as Amazon Connect Contact Lens, enabling post-contact summarization for increased productivity, and Amazon Q in Connect, which provides you the ability to easily understand customer intent and deliver accurate responses based on data source inputs, all in real-time. At the end of this session, learn how these easy-to-build configuration tools enrich your cloud center solution capabilities and improve customer experience.

Speaker: Joely Huang, Senior Cloud Architect, AWS Professional Services
Duration: 30mins

Build a generative AI-powered chatbot with Amazon Bedrock, Langchain, RAG and Streamlit (Level 300)
Many organizations are looking at using generative AI to build chatbot applications for business outcomes; including answering FAQs, scheduling, enhancing customer experiences through intuitive engagements and more. Join this session to uncover the key building blocks for a chatbot application powered by LLMs. We demonstrate how you can build chatbots using Amazon Bedrock APIs, Langchain, Streamlit, RAG and interact with high-performing foundation models such as Jurassic-2, Claude, Stable Diffusion, Command, and Amazon Titan. Understand how Amazon Bedrock’s new features such as knowledge bases and agents enable you to complete complex tasks for a wide range of use cases. We conclude by outlining the various tools you can use to build the chatbot best suited for your digital journey and business requirements.

Speaker: Aman Sharma, Senior Solutions Architect, AWS
Duration: 30mins

Deliver relevant, accurate, and customized responses with RAG and Amazon Bedrock (Level 200)
Generative AI applications can deliver better responses by incorporating organization-specific data with Retrieval Augmented Generation (RAG). However, implementing RAG requires specific skillset and time to configure connections to data sources, manage data ingestion workflows, and write custom code to manage the interactions between the foundation model (FM) and the data sources. This session covers how you can simplify the process with Amazon Bedrock knowledge bases and agents. From the user prompt, Amazon Bedrock automatically identifies data sources, retrieves the relevant information, and adds them to the prompt, thereby giving the FM more information to generate responses. At the end of this session, understand how these tools enable you to deliver more relevant, accurate, and customized responses based on your organization’s proprietary knowledge sources.

Speaker: Xin Chen, Senior Cloud Architect, AWS Professional Services
Duration: 30mins

Build a personalized shopping assistant with Agents for Amazon Bedrock and Amazon OpenSearch service (Level 200)
Do you wish your e-commerce website can quickly provide your users exactly what is relevant to them? Are you looking to leverage your existing systems such as databases, and microservices to create a personalized shopping assistant? In this session, we explain how to build personalized search experiences using Amazon OpenSearch Service and Amazon Bedrock to enhance your customer experience and improve conversion rates. We share the key reference architecture and demonstrate how this solution enables your customers to quickly locate the relevant products with the requested features. Understand how the agent can compare product information, and provide recommendations. We also explain how to use generative AI with your existing systems to build user-friendly sites, and provide your customers with prompt and precise search results.

Speaker: Tuan Huynh, Senior Cloud Architect, AWS Professional Services
Duration: 30mins

Unlock insights from your structured data using generative AI and analytics (Level 200)
Users from several organizations often rely on their technical teams to build the dashboards for insights needed for critical decision making. But the solutions built in-house at times require these users to know some SQL/Python languages in order to customize the dashboards to meet their requirements, which may delay the decision-making process. Join this session to learn how to build a simple dashboard using natural language and code generation capabilities of foundation models with Amazon Bedrock and AWS Glue Data Catalog. We explain how this easy-to-build solution provides users the ability to ask the questions about factual information, without requiring in-depth knowledge of where the data is located, structure of data, and programming languages. We also demonstrate how to access key insights from the data quickly using plain English language.

Speaker: Kamal Manchanda, Solutions Architect, AWS India
Duration: 30mins

Optimize your live video streaming experience with AI/ML (Level 200)
Live video streaming usage has been growing at exponential rate, and many video streaming content providers are looking to deliver real-time viewer engagement, and improve user experience. In this session, we showcase how to run high-quality, low-latency, and resilient live streams with AWS edge solutions. Find out how to optimize user experience in key areas such as content creation with generative AI on AWS, content moderation by leveraging Amazon Rekognition and content recommendations using Amazon Personalize.

Speakers: 
Thomas Sauvage, Senior Edge Services Go-to-Market Specialist, AWS
Christer Whitehorn, Principal Media Edge Services Specialist Solutions Architect, AWS

Duration: 30mins

Build a custom real-time multi-object tracking solution at the edge (Level 200)
Multi-object tracking (MOT) is a computer vision task that involves tracking movements of multiple objects over time in a video sequence. MOT can be applied in numerous practical use cases such as crowd monitoring, traffic management, security surveillance, production cycles, safety monitoring, and navigation as it offers the ability to analyze physical performance in real-time. Join this session as we demonstrate how to train a custom MOT model in Amazon SageMaker and prepare your model for deployment to the edge with AWS IoT Greengrass. We discuss key practical considerations when building a real-time tracking solution for the edge.

Speaker: Derrick Choo, Senior Solutions Architect, AWS
Duration: 30mins

Get started with Amazon SageMaker in minutes (Level 200)
Amazon SageMaker provides builders the ability to build, train, and deploy ML models for any use case with fully-managed infrastructure, tools, and workflows. In this session, find out how Amazon SageMaker takes away the heavy lifting, enabling high performance and low-cost ML. Explore the technical details of each of the modules of Amazon SageMaker, and the capabilities of the platform. Find out how to build generative AI applications with Amazon SageMaker to accelerate scalable, reliable, and secure foundation model (FM) development. The session features best practices to deploy open-source large language models (LLMs) using Amazon SageMaker and how to access LLMs via Amazon SageMaker Jumpstart. The session concludes with key takeaways with how you can analyze, evaluate, test, retrain, and deploy FMs, helping you to quickly get started.

Speaker: Pauline Kelly, Solutions Architect, AWS
Duration: 30mins

Improving productivity - Harness the power of generative AI foundation models on Amazon SageMaker Canvas (Level 200)
Machine learning can enable organizations of various industries and sizes to solve business challenges and achieve better outcomes. However, many are struggling to implement ML effectively throughout the organization, beyond their technical users. Join this session to discover how to use Amazon SageMaker Canvas to complete the ML lifecycle — from preparing data and creating models to generating predictions — without writing a single line of code. Learn how both technical and non-technical users in your organization can utilize ready-to-use models or create your own to gain insights from your data and ML models with Amazon SageMaker Canvas. We will explain how to access open-source and Amazon LLMs on Amazon SageMaker Canvas through a single interface, and provide guidance on how to simply prompt the models for assistance with tasks such as generating content, summarizing, categorizing documents, and answering questions with just a few clicks. The session also covers how to ask specific questions targeting your dataset and obtain answers to enhance your productivity, all without the need for an ML skillset.

Speaker: Sahil Verma, Senior Solutions Architect, AWS India
Duration: 30mins

Prepare ML data faster and at scale with Amazon SageMaker (Level 200)
Data preparation for ML can be challenging because it requires extracting, normalizing data and performing feature engineering which can be time-consuming. This session covers how you can simplify the process of data preparation, feature engineering, and complete each step of the data preparation workflow, including data selection, cleansing, exploration, bias detection, and visualization, from a single visual interface with Amazon SageMaker Data Wrangler. Find out how to reduce the time it takes to aggregate and prepare data for ML from weeks to minutes. We also share how simplify data preparation at multiple stages in Retrieval Augmented Generation (RAG) models with Amazon SageMaker Data Wrangler.

Speaker: Ben Friebe, Senior ISV Solutions Architect, AWS

Optimize training of foundation models on Amazon SageMaker (Level 200)
Training machine learning models at scale often requires a significant amount of resources, time, and investment. In this session, we explain how to leverage Amazon SageMaker to train and tune machine learning (ML) models without the need to manage infrastructure. We walk through the various strategies to train large language models in a cost-effective and performant manner using Amazon SageMaker. Learn how to optimize training to minimize cost and achieve high performance with distributed training approaches, smart sifting, Amazon SageMaker HyperPod, AWS Trainium, and other optimization methods. The session concludes with tips and best practices for optimizing training to minimize cost and achieve high performance.

Speakers: 
Gaurav Singh, Senior Solutions Architect, AWS India
Smiti Guru, Senior Solutions Architect, AWS India

Duration: 30mins

Building NLP models with Amazon SageMaker (Level 300)
Organizations have to manage massive volumes of voice and text data from various communication channels. Some use NLP to automatically process this data, analyze the intent or sentiment in the messages, and respond in real-time to human communication. But NLP models often consist of hundreds of millions of model parameters. Thus building, training, and optimizing them may require a lot of time, resources, and skills. This session outlines how Amazon SageMaker helps you to quickly build and train NLP models. We share the different distributed training and inference for large language models on Amazon SageMaker for use cases such as sentiment analysis, text summarization, and text classification.

Speaker: Tapan Hoskeri, Principal Solutions Architect, AWS India
Duration: 30mins

Implement MLOps practices with Amazon SageMaker (Level 200)
Implementing the right MLOps practices enables builders to collaborate effectively in preparing, building, training, deploying, and managing models at scale. In this session, we will showcase the MLOps features in Amazon SageMaker that assist in provisioning consistent model development environments, automating ML workflows, implementing CI/CD pipelines for ML, monitoring models in production, and standardizing model governance capabilities. We will demonstrate how to apply these MLOps practices using Amazon SageMaker features — such as SageMaker projects, SageMaker Pipelines, SageMaker Model Registry, and SageMaker Model Monitor — to quickly deliver high-performance production ML models at scale.

Speakers: 
Gaurav Singh, Senior Solutions Architect, AWS India
Smiti Guru, Senior Solutions Architect, AWS India

Duration: 30mins

Real-world MLOps for batch inference with model monitoring using Amazon SageMaker (Level 300)
Maintaining ML workflows in production is not easy, as there is a need to create continuous integration and continuous delivery (CI/CD) pipelines for ML code and models, manage model versioning, and monitor for data and concept drift, as well as model retraining. In addition, there is a manual approval process to ensure new versions of the model satisfy both performance and compliance requirements. Join this session to learn how to create ML workflows for batch inference to automate key steps such as job scheduling, model monitoring, retraining, and registration with Amazon SageMaker. Discover the best practices in MLOps and how to integrate them with your existing CI/CD and IaC (Infrastructure as Code) tools. We will share key approaches you can use to mitigate challenges and common pitfalls when adopting MLOps practices. The session concludes with guidance on how to reduce the complexities and costs associated with running and maintaining batch inference workloads in production.

Speaker: Indrajit Ghosalkar, Senior Solutions Architect, AWS
Duration: 30mins

Builders guide to AI/ML on AWS (Level 300)
Join this session to learn the strategies on how to select and leverage machine learning, and generative AI solutions on AWS when developing your software applications. We explain where traditional ML solutions including Amazon Comprehend and Amazon Personalize are best suited for select use cases, and when to apply generative models via AWS Bedrock. We then showcase how Amazon CodeWhisperer enables you to accelerate your overall software development process. This session also features a demo on how to generate code suggestions ranging from snippets to full functions in real-time in the IDE based on comments and existing codes. At the end of this session, understand when and how to utilize the right ML and generative AI solutions for your applications.

Speaker: Matt Coles, Principal Engineer, AWS
Duration: 30mins

Application development cookbook: Recipes for building applications with ML and generative AI (Level 300)
As the popularity of generative AI applications increases, so does the number of ways builders and developers can harness these capabilities and integrate them into their applications. Did you know that you can build applications with generative AI capabilities in languages other than Python? Are you aware that you can call Large Language Models (LLMs) from services such as AWS Lambda and AWS AppSync? Did you know there are reference codes available that you can leverage to build your applications more efficiently? Join this session to learn about new approaches to building applications, including an example application written in C#. Discover how to add generative AI capabilities such as Vector Embeddings and LangChain with Amazon Bedrock to your applications. We will also explore various options for calling Amazon Bedrock from other AWS services. By the end of this session, you will learn about the tools available to improve productivity and enhance user experience.

Speaker: Derek Bingham, Senior Developer Advocate, AWS
Duration: 30mins

Enrich and turbo charge your generative AI applications with visual workflow (Level 300)
Developers and builders are constantly looking at how to accelerate the development of generative AI applications to deliver rapid innovation, without requiring extensive investment in time, budgets and resources. Join this session to learn how to use AWS Step Functions, a visual workflow solution, to build distributed applications, automate processes, and create data and ML pipelines. We demonstrate the use of AWS Step Functions for orchestrating generative AI applications with Amazon Bedrock foundation models. At the end of the session, learn how these key building blocks enable you to compose LLMs and integrate with various AWS services into production-grade workflows.

Speaker: Donnie Prakoso, Principal Developer Advocate, AWS
Duration: 30mins