Design prompts for JSON-only LLM responses

Q: Design prompts for JSON-only LLM responses

This question evaluates a candidate's skills in prompt engineering, API and schema design, and robust output validation for large language models, with emphasis on producing strict JSON-only responses.

Q: How do I approach ML System Design interview questions?

ML System Design questions require understanding of core concepts and practice. PracHub provides solutions with explanations to help you master ml system design interviews.

Q: What difficulty level is this interview question?

This is a medium difficulty ML System Design question, commonly asked during Technical Screen rounds at Intuit.

Q: What role is this question designed for?

This question is commonly asked for Software Engineer candidates at Intuit during technical interviews.

Question

Design an LLM request that returns strict JSON only

Context

You are designing a backend call to an LLM for a technical screen (System Design & Engineering, Software Engineer). The service must return valid JSON only—no prose, no Markdown, no code fences—so downstream code can parse and validate responses reliably.

Task

Describe how you would:

Structure message roles and instructions so the model returns only JSON.
Constrain outputs using either a JSON Schema or function/tool calling.
Choose parameters (e.g., temperature, top_p, max tokens) to reduce variability.
Enforce correctness at runtime (schema validation, parsing, retries with repair prompts, and/or few-shot JSON examples).
Provide a minimal example of the request and the expected JSON shape.
Explain how you would evaluate whether the output meets the “ideal answer” criteria and how you would iteratively improve prompts and guards to increase reliability.

Design prompts for JSON-only LLM responses

Quick Overview

Design prompts for JSON-only LLM responses

Design an LLM request that returns strict JSON only

Context

Task

Submit Your Answer to Earn 20XP