Integrate duohub’s memory retrieval capabilities with AWS Lambda functions to query your knowledge graph through serverless functions.
Quick Start
Here’s our production-ready TypeScript implementation:
import { Handler } from "aws-lambda";
import axios from "axios";
interface LambdaEvent {
query: string;
memoryID?: string;
}
interface MemoryRetrieverResponse {
payload: string;
facts: Array<{
text: string;
relevance: number;
}>;
}
export const handler: Handler<LambdaEvent, LambdaResponse> = async (event) => {
try {
const { query, memoryID } = event;
const params = new URLSearchParams({
query,
...(memoryID && { memoryID }),
assisted: "true",
facts: "true",
});
const response = await axios.get<MemoryRetrieverResponse>(
`https://api.duohub.ai/memory/?${params.toString()}`,
{
headers: {
"X-API-KEY": process.env.API_KEY as string,
"Content-Type": "application/json",
},
}
);
return {
success: true,
message: "Query executed successfully.",
answer: response.data.payload,
facts: response.data.facts,
};
} catch (err) {
console.error("Error querying memory retriever:", err);
return {
success: false,
message: `Error: ${err instanceof Error ? err.message : String(err)}`,
answer: "",
facts: [],
};
}
};
Prerequisites
- AWS Account with Lambda access
- Node.js 18 or later
- duohub API key
- AWS CLI configured (if deploying manually)
Setup Guide
1. Environment Variables
Set up your Lambda environment variable:
API_KEY=your-duohub-api-key
2. Dependencies
Required npm packages:
{
"dependencies": {
"aws-lambda": "^1.0.7",
"axios": "^1.7.7"
}
}
3. Deployment Options
Option A: Serverless Framework (Recommended)
Create a serverless.yml:
service: duohub-memory-service
provider:
name: aws
runtime: nodejs18.x
environment:
API_KEY: ${env:API_KEY}
functions:
memoryQuery:
handler: index.handler
events:
- http:
path: query
method: post
Deploy with:
Option B: Manual Deployment
- Build the TypeScript:
- Create a ZIP with the compiled code and node_modules:
zip -r function.zip dist/ node_modules/
- Upload to AWS Lambda through the console or CLI
4. IAM Configuration
Minimum required IAM policy:
{
"Version": "2012-10-17",
"Statement": [
{
"Effect": "Allow",
"Action": [
"logs:CreateLogGroup",
"logs:CreateLogStream",
"logs:PutLogEvents"
],
"Resource": "arn:aws:logs:*:*:*"
}
]
}
Usage
{
"query": string, // Required: Your query
"memoryID": string // Optional: Specific memory ID
}
{
"success": boolean,
"message": string,
"answer": string,
"facts": Array<{
"text": string,
"relevance": number
}>
}
Testing
Test your deployed function with curl:
curl -X POST https://your-lambda-url/query \
-H "Content-Type: application/json" \
-d '{"query": "your query", "memoryID": "optional-memory-id"}'
Best Practices
-
Error Handling
- Implement retries for network failures
- Add detailed logging
- Return meaningful error messages
-
Security
- Store API key in AWS Secrets Manager
- Use IAM roles with minimum required permissions
- Enable AWS X-Ray for request tracing
-
Performance
- Keep dependencies minimal
- Use connection pooling in axios
- Consider implementing caching if appropriate
-
Monitoring
- Set up CloudWatch alarms
- Monitor error rates and latency
- Set up logging for debugging
Troubleshooting
Common issues and solutions:
-
Timeout Issues
- Increase Lambda timeout in configuration
- Check duohub API response times
- Implement request timeouts in axios
-
Memory Issues
- Increase Lambda memory allocation
- Check for memory leaks
- Monitor memory usage patterns
-
Authorization Errors
