Generative AI on AWS
Prashant Singh
Solutions Architect Manager
Amazon Web Services
© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Agenda
• Introduction to Generative AI
• ReInventing with Generative AI on AWS
• Amazon Bedrock AI Stylist Demo & Architecture Walkthrough
• Build a GenAI based code generator using Amazon Bedrock
© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Innovation can
GENERATIVE AI
© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Question: What is generative AI?
• Creates new content and ideas, including conversations,
stories, images, videos, and music
• Powered by large models that are pretrained on vast corpora
of data and commonly referred to as foundation models (FMs)
BLOOM
© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved. © 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
AI
MACHINE SIMPLE SIMPLE
LEARNING INPUTS OUTPUTS
DEEP COMPLEX SIMPLE
LEARNING
INPUTS OUTPUTS
FOUNDATION COMPLEX COMPLEX
MODELS
INPUTS OUTPUTS
© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
How does a foundation model work?
Data Foundation
Text
model
Images
Speech
Structured data
3D signals
Fine-tune for
Gather data at Pre-train Evaluate
*can take weeks specific tasks
scale model
or even months and domains
*can take hours
© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Generative AI is powered by
foundation models
Pretrained on vast amounts of
unstructured data
Contain large number of parameters that make
them capable of learning complex concepts
Can be applied in a wide range of contexts
Customize FMs using your data for domain
specific tasks
© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
ML innovation is in Amazon’s DNA
of Alexa
sold every day interactions each week technology in airports,
on Amazon.com stadiums and more
© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Generative AI Stack
© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Generative AI Stack
APPLICATIONS THAT LEVERAGE LLMs AND OTHER
FMs
TOOLS TO BUILD WITH LLMs AND OTHER FMs
© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
CG1 G2 P2 G3 P3 G4 P4 G5 G5g P5
NVIDIA Tesla NVIDIA GRID NVIDIA NVIDIA NVIDIA V100 NVIDIA T4 NVIDIA A100 NVIDIA A10G NVIDIA T4G NVIDIA H100
M2050 “Fermi” GK104 “Kepler” K80 Tesla M60 Tensor Core Tensor Core Tensor Core Tensor Core Tensor Core Tensor Core
GPUs GPUs GPUs GPUs GPUs GPUs GPUs GPUs GPUs GPUs
Innovating at the silicon level
AWS AWS
HIGHER THROUGHPUT LOWER LATENCY
Generative AI Stack
APPLICATIONS THAT LEVERAGE LLMs AND OTHER FMs
INFRASTRUCTURE FOR FM TRAINING AND INFERENCE
GPUs Trainium Inferentia SageMaker
UltraClusters EFA EC2 Capacity Blocks Nitro Neuron
Choice of industry-leading FMs from AI21
Labs, Amazon, Anthropic, Cohere, Meta,
and Stability AI
Amazon Customize FMs using your
The easiest way to build and scale organization’s data
generative AI applications with LLMs
and other FMs
Enterprise-grade security and privacy
Amazon
Broad choice of models
JURASSIC-2 AMAZON TITAN CLAUDE COMMAND + EMBED LLAMA 2 STABLE DIFFUSION XL
Generative AI Stack
APPLICATIONS THAT LEVERAGE LLMs AND OTHER
FMs
Amazon Bedrock
Guardrails Agents Customization Capabilities
INFRASTRUCTURE FOR FM TRAINING AND INFERENCE
GPUs Trainium Inferentia SageMaker
UltraClusters EFA EC2 Capacity Blocks Nitro Neuron
Generative AI Stack
TOOLS TO BUILD WITH LLMs AND OTHER FMs
Amazon Bedrock
Guardrails Agents Customization Capabilities
INFRASTRUCTURE FOR FM TRAINING AND INFERENCE
GPUs Trainium Inferentia SageMaker
UltraClusters EFA EC2 Capacity Blocks Nitro Neuron
Provides interactive answers, solves problems,
generates content, and takes action
NEW
Understands your company information,
Amazon Q code, and systems
A generative AI-powered assistant for
work that is tailored to your business Personalizes interactions based on your role
and permissions
AVAILABLE IN PREVIEW
Built to be secure and private
is
AMAZON Q AMAZON Q IN AMAZON QUICKSIGHT
AMAZON Q AMAZON Q IN AMAZON CONNECT
Generative AI Stack
Amazon Q Amazon Q in Amazon Q in Amazon
Amazon QuickSight Amazon Connect CodeWhisperer
Amazon Bedrock
Guardrails Agents Customization Capabilities
GPUs Trainium Inferentia SageMaker
UltraClusters EFA EC2 Capacity Blocks Nitro Neuron
https://aistylist.awsplayer.com/
AiStylist Solution Components (RAG & Embeddings)
Generate Embeddings
Request Orchestration
© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
AiStylist Solution Architecture
© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
Code Generator
AWS Cloud
Amazon API Gateway Amazon Bedrock
AWS Lambda
User
Learning Resources
Hands-on Course on Generative AI with
Amazon Bedrock Workshop Large Language Models
AWS Learning Needs Analysis: Learn more about AWS Skill Builder:
Build a data-driven plan to accelerate learning
Additional Reading
REACT: Synergizing Reasoning and Acting in Language Models
Chain-of-Thought Prompting Elicits Reasoning in Large Language Models
Attention is All You Need
Session Feedback
Please provide your valuable
feedback by Scanning the
QRCode.
This helps us raise the bar on
our engagements with you.
© 2023, Amazon Web Services, Inc. or its affiliates. All rights reserved.
