AI Data Engineer Needed to Fix LLM Data Processing Issue

We have a small AI data pipeline that processes structured JSON records through an LLM and saves the results to PostgreSQL. Most of it is working, but some records are returning inconsistent fields, missing values, and formatting differences that are causing problems downstream. We’re looking for an experienced AI/data engineer to review the current Python processing logic, identify the cause, and make the output consistent and reliable. The task should involve reviewing the prompt/output schema, improving validation and error handling, and testing the fix against a small sample dataset. Stack: Python, OpenAI API, JSON, PostgreSQL. This is a focused 1–2 hour task, not a long-term project. Please briefly mention similar LLM structured-output or data pipeline work you’ve done.