AI Automation Fix Debug & Repair Existing LLM Workflow

I have an existing AI automation (LLM-powered workflow) that's partially working but needs an experienced hand to fix and stabilise it. This is a defined scope, not an open-ended build. What I need: Diagnose why the AI workflow is producing inconsistent/incorrect outputs Fix prompt logic, API calls, or integration issues as needed Ensure reliable end-to-end runs with proper error handling Short handover note on what was changed and why Likely involves: OpenAI/Anthropic API calls (chat completions, structured outputs) Prompt debugging and output formatting (JSON parsing issues, etc.) Possibly n8n / Make / Python scripts depending on the stack Requirements: Proven experience building or fixing LLM integrations (show me one example) Understands API rate limits, token handling, and structured outputs Can start within 24-48 hours Clear communicator — I want short updates, not silence To apply: One short paragraph — describe a similar AI fix you've done and how you approached it. Applications that just paste a generic profile get skipped. Quick, quality work here opens the door to ongoing AI automation projects.