@aeye/aws - AWS Bedrock Provider

AWS Bedrock integration for the @aeye framework, providing seamless access to Claude, Llama, Mistral, Cohere, Stability AI, and Amazon Titan models.

Features

• 🤖 Multi-Model Support - Access to all AWS Bedrock foundation models
• 🔐 Automatic Authentication - Uses AWS SDK credential chain for seamless authentication
• 🌊 Streaming - Full streaming support for real-time responses
• 🎨 Image Generation - Stability AI models for image creation
• 📊 Embeddings - Amazon Titan embeddings for semantic search
• 🔧 Type-Safe - Full TypeScript support with type inference
• ⚡ Regional Flexibility - Deploy across any AWS region with Bedrock

Supported Models

Chat Models

• Anthropic Claude - Claude 3 Opus, Sonnet, Haiku, and Claude 3.5 Sonnet
• Meta Llama - Llama 2, Llama 3, Llama 3.1, Llama 3.2
• Mistral AI - Mistral 7B, Mixtral 8x7B, Mistral Large
• Cohere - Command, Command Light, Command R, Command R+
• AI21 Labs - Jurassic-2 models

Image Models

• Stability AI - Stable Diffusion XL, Stable Diffusion 3

Embedding Models

• Amazon Titan - Titan Text Embeddings v1 and v2

Installation

npm install @aeye/aws @aeye/ai @aeye/core zod

Prerequisites

AWS Credentials

The provider uses the AWS SDK v3 credential chain to automatically discover credentials:

1. Environment variables - AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY, AWS_SESSION_TOKEN
2. Shared credentials file - ~/.aws/credentials
3. IAM roles - When running on EC2, ECS, Lambda, etc.

AWS Bedrock Access

Ensure you have:

1. Access to AWS Bedrock in your target region
2. Permissions to invoke foundation models
3. Model access enabled in the AWS Bedrock console

Quick Start

Basic Chat Completion

import { AI } from '@aeye/ai';
import { AWSBedrockProvider } from '@aeye/aws';

// Create provider (credentials picked up automatically)
const aws = new AWSBedrockProvider({
  region: 'us-east-1',
});

// Create AI instance
const ai = AI.with()
  .providers({ aws })
  .create();

// Chat with Claude
const response = await ai.chat.get([
  { role: 'user', content: 'What is AWS Bedrock?' }
], {
  model: 'anthropic.claude-3-sonnet-20240229-v1:0'
});

console.log(response.content);

Streaming Responses

// Stream responses in real-time
for await (const chunk of ai.chat.stream([
  { role: 'user', content: 'Write a poem about AI' }
], {
  model: 'anthropic.claude-3-haiku-20240307-v1:0'
})) {
  if (chunk.content) {
    process.stdout.write(chunk.content);
  }
}

Image Generation

import { AWSBedrockProvider } from '@aeye/aws';

const aws = new AWSBedrockProvider({
  region: 'us-east-1',
  defaultModels: {
    imageGenerate: 'stability.stable-diffusion-xl-v1',
  },
});

const ai = AI.with()
  .providers({ aws })
  .create();

const result = await ai.image.generate({
  prompt: 'A serene mountain landscape at sunset',
  size: { width: 1024, height: 1024 },
});

console.log('Image URL:', result.images[0].url);

Text Embeddings

const aws = new AWSBedrockProvider({
  region: 'us-east-1',
  defaultModels: {
    embedding: 'amazon.titan-embed-text-v1',
  },
});

const ai = AI.with()
  .providers({ aws })
  .create();

const embeddings = await ai.embed.get({
  input: 'AWS Bedrock is a fully managed service',
  model: 'amazon.titan-embed-text-v1',
});

console.log('Embedding dimensions:', embeddings.embeddings[0].length);

Configuration

Provider Configuration

import { AWSBedrockProvider, type AWSBedrockConfig } from '@aeye/aws';

const config: AWSBedrockConfig = {
  // AWS region (defaults to AWS_REGION env var or 'us-east-1')
  region: 'us-west-2',
  
  // Optional: Explicit credentials (if not using default chain)
  credentials: {
    accessKeyId: process.env.AWS_ACCESS_KEY_ID!,
    secretAccessKey: process.env.AWS_SECRET_ACCESS_KEY!,
    sessionToken: process.env.AWS_SESSION_TOKEN, // Optional
  },
  
  // Optional: Default models for different capabilities
  defaultModels: {
    chat: 'anthropic.claude-3-sonnet-20240229-v1:0',
    imageGenerate: 'stability.stable-diffusion-xl-v1',
    embedding: 'amazon.titan-embed-text-v2:0',
  },
};

const provider = new AWSBedrockProvider(config);

Using with Multiple Providers

import { AI } from '@aeye/ai';
import { AWSBedrockProvider } from '@aeye/aws';
import { OpenAIProvider } from '@aeye/openai';

const aws = new AWSBedrockProvider({ region: 'us-east-1' });
const openai = new OpenAIProvider({ apiKey: process.env.OPENAI_API_KEY! });

const ai = AI.with()
  .providers({ aws, openai })
  .create({
    // Automatic model selection across all providers
    defaultMetadata: {
      required: ['chat', 'streaming'],
      weights: {
        cost: 0.5,
        speed: 0.3,
        quality: 0.2,
      },
    },
  });

// AI instance will automatically select the best model across AWS and OpenAI
const response = await ai.chat.get([
  { role: 'user', content: 'Hello!' }
]);

Model IDs

Anthropic Claude Models

// Claude 3.5
'anthropic.claude-3-5-sonnet-20240620-v1:0'
'anthropic.claude-3-5-sonnet-20241022-v2:0'

// Claude 3
'anthropic.claude-3-opus-20240229-v1:0'
'anthropic.claude-3-sonnet-20240229-v1:0'
'anthropic.claude-3-haiku-20240307-v1:0'

// Claude 2.x
'anthropic.claude-v2:1'
'anthropic.claude-v2'
'anthropic.claude-instant-v1'

Meta Llama Models

'meta.llama3-2-1b-instruct-v1:0'
'meta.llama3-2-3b-instruct-v1:0'
'meta.llama3-2-11b-instruct-v1:0'
'meta.llama3-2-90b-instruct-v1:0'
'meta.llama3-1-8b-instruct-v1:0'
'meta.llama3-1-70b-instruct-v1:0'
'meta.llama3-1-405b-instruct-v1:0'
'meta.llama3-8b-instruct-v1:0'
'meta.llama3-70b-instruct-v1:0'
'meta.llama2-13b-chat-v1'
'meta.llama2-70b-chat-v1'

Mistral AI Models

'mistral.mistral-7b-instruct-v0:2'
'mistral.mixtral-8x7b-instruct-v0:1'
'mistral.mistral-large-2402-v1:0'
'mistral.mistral-large-2407-v1:0'

Cohere Models

'cohere.command-text-v14'
'cohere.command-light-text-v14'
'cohere.command-r-v1:0'
'cohere.command-r-plus-v1:0'

Stability AI Models

'stability.stable-diffusion-xl-v1'
'stability.sd3-large-v1:0'

Amazon Titan Models

// Embeddings
'amazon.titan-embed-text-v1'
'amazon.titan-embed-text-v2:0'

// Image Generation
'amazon.titan-image-generator-v1'
'amazon.titan-image-generator-v2:0'

Advanced Features

Vision with Claude 3

const response = await ai.chat.get([
  {
    role: 'user',
    content: [
      { type: 'text', content: 'What do you see in this image?' },
      { type: 'image', content: 'https://example.com/image.jpg' },
    ],
  },
], {
  model: 'anthropic.claude-3-sonnet-20240229-v1:0',
});

Custom Request Parameters

const response = await ai.chat.get([
  { role: 'user', content: 'Hello!' }
], {
  model: 'anthropic.claude-3-haiku-20240307-v1:0',
  temperature: 0.7,
  maxTokens: 2048,
  topP: 0.9,
  stop: ['Human:', 'Assistant:'],
});

Multi-Region Setup

// Primary region
const awsEast = new AWSBedrockProvider({
  region: 'us-east-1',
});

// Fallback region
const awsWest = new AWSBedrockProvider({
  region: 'us-west-2',
});

const ai = AI.with()
  .providers({ awsEast, awsWest })
  .create();

Error Handling

import {
  AWSError,
  AWSAuthError,
  AWSRateLimitError,
  AWSQuotaError,
  AWSContextWindowError,
} from '@aeye/aws';

try {
  const response = await ai.chat.get([
    { role: 'user', content: 'Hello!' }
  ]);
} catch (error) {
  if (error instanceof AWSAuthError) {
    console.error('Authentication failed. Check your AWS credentials.');
  } else if (error instanceof AWSRateLimitError) {
    console.error('Rate limit exceeded. Retry after:', error.retryAfter);
  } else if (error instanceof AWSQuotaError) {
    console.error('AWS quota exceeded. Request quota increase.');
  } else if (error instanceof AWSContextWindowError) {
    console.error('Input too long for model context window.');
  } else if (error instanceof AWSError) {
    console.error('AWS Bedrock error:', error.message);
  }
}

Health Check

const isHealthy = await aws.checkHealth();
console.log('AWS Bedrock is accessible:', isHealthy);

List Available Models

const models = await aws.listModels();
console.log('Available models:', models.map(m => m.id));

// Filter by capability
const chatModels = models.filter(m => m.capabilities.has('chat'));
const imageModels = models.filter(m => m.capabilities.has('image'));
const embeddingModels = models.filter(m => m.capabilities.has('embedding'));

Best Practices

1. Use IAM Roles - When running on AWS infrastructure (EC2, ECS, Lambda), use IAM roles instead of hardcoded credentials
2. Region Selection - Choose regions close to your users for lower latency
3. Model Selection - Use smaller models (Haiku, 7B) for faster, cheaper responses when appropriate
4. Streaming - Enable streaming for better user experience with long responses
5. Error Handling - Implement proper error handling and retry logic for production use
6. Cost Management - Monitor usage through AWS Cost Explorer and set up billing alerts

Environment Variables

# AWS credentials (if not using IAM roles)
AWS_ACCESS_KEY_ID=your_access_key
AWS_SECRET_ACCESS_KEY=your_secret_key
AWS_SESSION_TOKEN=your_session_token  # Optional

# AWS region
AWS_REGION=us-east-1

Links

• AWS Bedrock Documentation
• Model IDs Reference
• @aeye Framework
• AWS SDK for JavaScript

License

GPL-3.0 - See LICENSE file for details.

Contributing

Issues and pull requests are welcome! See the main repository for contribution guidelines.