Browse categories
Explore
Fiverr Pro
English
$
USD
I will fix inconsistent json outputs in your python llm API workflow
Mhamed A
About this gig
Is your Python LLM workflow returning malformed JSON, missing fields, wrong types, or inconsistent structured output?
I will diagnose and harden an existing LLM API workflow so bad model output is caught before it breaks the next step. Depending on your current setup, I can add JSON/schema validation, explicit failure states, bounded retries when appropriate, clearer error handling, logging, and focused tests.
Best fit:
- OpenAI/GPT, DeepSeek, Claude, or similar LLM API workflows
- Python applications that depend on reliable structured output
- Existing code that intermittently fails because model responses are inconsistent
You will receive clean source-code changes and a concise explanation of what was fixed. I focus on a small, testable scope rather than rebuilding your whole application.
Public code proof:
https://github.com/zongsm-cmyk/llm-reliability-rescue-demo
The public proof includes a FastAPI/Pydantic reliability layer and 11/11 regression tests.
Please message me with a redacted code snippet, sample output/schema, or error log so I can confirm the scope before we start.
Get to know Mhamed A
Mhamed A
AI Integration Automation Developer
- FromMorocco
- Member sinceSep 2026
Languages
Arabic, English, French
I am a Python and AI automation developer focused on building reliable LLM workflows with structured outputs and rigorous validation. I specialize in workflow engineering, including provider routing, failure handling, and reproducible testing. I also have extensive experience in Arabic content quality and AI output evaluation.
FAQ
What do you need from me before you start?
Please send a redacted code snippet, the expected JSON/schema if you have one, one or two failing model responses, and the relevant error log. Do not send passwords or production secrets.
Will you rebuild my whole application?
No. This Gig is intentionally scoped around an existing Python LLM workflow. I focus on the structured-output failure, validation, retry/failure handling, and the agreed surrounding code.
Which LLM APIs can you work with?
The strongest fit is Python workflows using GPT/OpenAI, DeepSeek, or similar LLM APIs that return structured data. If you use another provider, send the current code/API shape first and I will confirm fit.
