AI Engineering › 13 of 14

Typed Data with Pydantic

Validate LLM outputs and API payloads.

python

from pydantic import BaseModel, Field

class Sentiment(BaseModel):
    label: str = Field(pattern="^(pos|neg|neutral)$")
    confidence: float = Field(ge=0, le=1)

result = Sentiment.model_validate_json(llm_output)
print(result.label, result.confidence)

⚡ Quick check

✓ BaseModel for schemas
✓ model_validate_json()
✓ Field(ge=, le=) constraints