Generation of JSONL + parsing of the result for OpenAI Batch API

There is a CSV with approximately 55,000 companies (ID, name, location) and a clearly configured and tested prompt for classification into 14 industry categories. Attached is a screenshot of part of the CSV file and a text file with the prompt.

Required:

1. Based on my CSV and prompt:

   • split the data into batches of 50 companies;

   • generate a file requests.jsonl in OpenAI Batch API format for POST /v1/responses

   • each line of JSONL must contain my prompt plus 50 lines of companies.

2. I will upload requests.jsonl to OpenAI myself, run the Batch, and upload the results file results.jsonl.

3. After that, it is necessary to:

   • parse results.jsonl;

   • extract from the model responses pairs ID,ShortLabel (where ShortLabel is one of the 14 codes or Unknown);

   • return the final CSV with columns ID,ShortLabel;

   • separately mark problematic or unreadable lines, if any.

Important conditions:

• Access to my OpenAI API key is not needed, I will run the Batch myself.

• The content of the prompt cannot be changed, only minimal cosmetic formatting is allowed.

• This is a one-time task: a single correct run on all 55,000 lines is sufficient.

Desired skills:

• Experience with Python or Node.js.

• Ability to work with JSONL files.

• Real experience with OpenAI API is preferred, ideally with Batch API.

In your response, please specify:

• Which language you will use (Python or Node).

• If you have experience specifically with OpenAI API / Batch.