Get JSON output from an LLM and validate it against a schema using AWS Lambda durable functions
This pattern demonstrates structured output extraction with AWS Lambda durable functions. The LLM response is validated against a schema with automatic retry on validation errors.
The example uses an LLM to extract contact information from unstructured text and validates it using Pydantic (Python) or Zod (TypeScript).
Benefits: Automatic retry on validation failures without manual error handling. Schema enforcement ensures data quality. Checkpointing prevents re-extraction on transient failures.
Select language:
Usage Example with Python:
Structured output with Lambda durable functions using Python and Pydantic.
import json
import re
import boto3
from aws_durable_execution_sdk_python import DurableContext, durable_execution
from pydantic import BaseModel
MODEL_ID = "us.amazon.nova-lite-v1:0"
bedrock = boto3.client("bedrock-runtime")
def converse(model_id: str, prompt: str) -> str:
response = bedrock.converse(
modelId=model_id,
messages=[{"role": "user", "content": [{"text": prompt}]}],
)
return response["output"]["message"]["content"][0]["text"]
class ExtractedContact(BaseModel):
name: str
email: str
company: str
def extract_contact(text: str) -> dict:
raw = converse(
MODEL_ID,
f'Extract contact info as JSON with keys "name", "email", "company": {text}',
)
match = re.search(r"\{[^}]+\}", raw)
if not match:
raise ValueError("No JSON found in response")
data = json.loads(match.group())
return ExtractedContact(**data).model_dump()
@durable_execution
def handler(event: dict, context: DurableContext):
text = event.get("text", "John Smith from Acme Corp, email: john@acme.com")
return context.step(lambda _: extract_contact(text), "extract")
Created by:
